Optimisation et réduction de la dose d’irradiation au

scanner : aspects techniques et impact en pratique

clinique courante
Alban Gervaise

To cite this version:

Alban Gervaise. Optimisation et réduction de la dose d’irradiation au scanner : aspects techniques et
impact en pratique clinique courante. Médecine humaine et pathologie. Université de Lorraine, 2016.
Français. �NNT : 2016LORR0136�. �tel-01491102�

HAL Id: tel-01491102

[Link]
Submitted on 16 Mar 2017

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 AVERTISSEMENT

Ce document est le fruit d'un long travail approuvé par le jury de

soutenance et mis à disposition de l'ensemble de la
communauté universitaire élargie.

Il est soumis à la propriété intellectuelle de l'auteur. Ceci

implique une obligation de citation et de référencement lors de
l’utilisation de ce document.

D'autre part, toute contrefaçon, plagiat, reproduction illicite

encourt une poursuite pénale.

Contact : ddoc-theses-contact@[Link]

LIENS

Code de la Propriété Intellectuelle. articles L 122. 4

Code de la Propriété Intellectuelle. articles L 335.2- L 335.10
[Link]
[Link]
 Ecole Doctorale BioSE (Biologie-Santé-Environnement)

Thèse

Présentée et soutenue publiquement pour l’obtention du titre de

DOCTEUR DE L’UNIVERSITE DE LORRAINE

Mention : « Sciences de la Vie et de la Santé »

par Alban GERVAISE

« Optimisation et réduction de la dose d’irradiation au scanner :

aspects techniques et impact en pratique clinique courante »
03/10/2016

Membres du jury :

Rapporteurs :
Pr DUCOU LE POINTE Hubert PU-PH, HDR, Université Paris VI, France
Dr BRISSE Hervé Praticien spécialiste, HDR, Université Paris V, France

Examinateurs :
Pr DRAPE Jean-Luc PU-PH, HDR, Université Paris V, France
Pr BLUM Alain PU-PH, HDR, Université de Lorraine, Nancy, France
Directeur de thèse
Dr TEIXEIRA Pedro MCU-PH, Université de Lorraine, Nancy, France
Dr NOEL Alain Physicien médical, HDR, Université de Lorraine, Nancy, France

Membres invités :
Pr FOEHRENBACH Hervé Professeur agrégé du Val de Grâce, DCSSA, Paris, France
Dr TACK Denis Praticien spécialiste, Hôpital RHMS, Baudour, Belgique

-----------------------------------------------------------------------------------------------------------------------------------

INSERM U947 Laboratoire « Imagerie Adaptative Diagnostique et Interventionnelle »

CHU de Nancy-Brabois – Bâtiment Recherche – Rez-de-chaussée
rue du Morvan, 54511 Vandœuvre-lès-Nancy
 Remerciements

Cette thèse est le fruit de nombreuses années de travail et le résultat de nombreuses collaborations. Je tiens donc
à remercier tous ceux qui ont participé de près ou de loin à la réalisation de ces travaux de recherche.

En premier lieu, je remercie le Professeur Jacques Felblinger, directeur du laboratoire d’Imagerie Adaptative
Diagnostique et Interventionnelle du CHRU de Nancy, de m’avoir accueilli au sein du laboratoire et de m’avoir
toujours poussé à réaliser cette thèse. Même si ma présence au sein du laboratoire a été épisodique, j’en garde un
excellent souvenir. Je tiens également à remercier tous les membres du laboratoire qui m’ont aidé dans mon
travail. En particulier, je retiendrai ma collaboration fructueuse avec Emilien Micard et Gabriela Hossu. Que ce
soit pour l’informatique ou les statistiques, vous avez toujours été disponibles pour m’aider dans mes différents
projets.

Je remercie mon directeur de thèse, le Professeur Alain Blum, chef du service d’imagerie Guilloz du CHRU
Nancy, de m’avoir donné l’opportunité de réaliser ce travail au sein de son service, de m’avoir accordé sa
confiance et de m’avoir permis d’utiliser les équipements modernes dont dispose son service.

Un grand merci au Docteur Pedro Teixeira, mon « presque » co-directeur de thèse pour ses conseils, sa bonne
humeur, sa disponibilité et ses compétences scientifiques indéniables. J’espère que nous aurons l’occasion de
pouvoir retravailler ensemble !

Je remercie le Professeur Drapé, président du jury de thèse, de l’attention et de l’intérêt portés à mon travail. Ses
remarques pertinentes et constructives me permettront d’améliorer mes futurs travaux. Je remercie le Professeur
Ducou-Lepointe et le Docteur Brisse d’avoir accepté d’être rapporteurs de mon jury de thèse. Leur expérience en
tant que radio-pédiatres et leur expertise en radio-protection ont été d’une grande valeur. Leurs remarques et
questions pointues sur mon travail me poussent à encore améliorer mes connaissances sur ce sujet. Je remercie le
Professeur Foehrenbach, ancien chef du service de médecine nucléaire de l’Hôpital du Val de Grâce et
spécialiste en radioprotection. Son expertise en matière de radioprotection des patients est un modèle pour moi.
J’espère pouvoir être à la hauteur des « grands anciens » qui ont fait la réputation du Service de Santé des
Armées en matière de radioprotection. Je remercie le Docteur Alain Noël d’avoir accepté d’être examinateur
dans mon jury de thèse et de m’avoir fait profiter de ses compétences en radio-physique médicale. Enfin, je
remercie le Docteur Denis Tack, tout d’abord pour m’avoir fait l’honneur de faire la route depuis la Belgique, et
surtout pour son expertise en radioprotection des patients et en optimisation de la dose au scanner. Son travail de
thèse de sciences a été un modèle pour moi depuis de nombreuses années et j’espère que mon travail a pu être à
la hauteur du sien.

Je remercie l’équipe médicale et paramédicale du service d’imagerie Guilloz. Cela a toujours été un plaisir pour
moi de venir dans ce service. Merci en particulier aux médecins qui ont bien voulu jouer le rôle de relecteur dans
mes différentes études : Sophie Lecocq, Benoit Osemont, Mathias Louis, Johny Wassel.
Un grand merci à l’équipe du Service d’Imagerie Médicale de l’HIA Legouest. Ce service restera à jamais mon
premier service de radiologie. J’y ai fait mes débuts et je me suis épanoui auprès de vous. Grâce à votre bonne
humeur collégiale, j’ai pu progresser et mener mes projets à bien. C’est avec émotion que je vous ai quittés pour
de nouvelles aventures parisiennes mais je ne vous oublierai jamais !

Merci aux différents co-auteurs de mes articles ainsi qu’à tous ceux qui ont travaillé dans l’ombre pour permettre
leur publication.

Enfin, et surtout, je remercie mon épouse pour son soutien indéfectible et sa compréhension. Tu as toujours cru
en moi et tu m’as toujours soutenu. Je te remercie infiniment ma chérie. Je t’aime. Tu m’as aussi donné un
superbe fils, Paul, qui est plein de vie ! Mais plus que n’importe quelle thèse, c’est notre famille qui est
importante à mes yeux !!! Quant à toi Paul, j’espère que plus tard tu seras fier de ton Papa comme je suis déjà
fier de toi.
 TABLE DES MATIERES

TABLE DES MATIERES ....................................................................................................1

ABREVIATIONS .................................................................................................................3
TABLE DES FIGURES .......................................................................................................5
TABLE DES TABLEAUX ...................................................................................................7
INTRODUCTION GENERALE..........................................................................................9
LISTE DES PUBLICATIONS ET COMMUNICATIONS ..............................................15
CHAPITRE 1 : CONTEXTE ET PROBLEMATIQUE ...................................................19
1- Contexte....................................................................................................................21
2- Problématique ...........................................................................................................25
3- Environnement de travail...........................................................................................36
CHAPITRE 2 : INFLUENCE DES FACTEURS COMPORTEMENTAUX ..................41
Article 1 : Evaluation des connaissances des prescripteurs de scanner en matière de
radioprotection des patients...............................................................................................47
Article 2 : Evaluation de l’intérêt de l’acquisition abdominopelvienne sans injection lors de
la réalisation d’un scanner corps entier chez un patient suspect de polytraumatisme..........57
Article 3 : Optimisation de la longueur d’acquisition des scanners réalisés pour colique
néphrétique : proposition d’une nouvelle méthode pour le placement de la limite supérieure
de l’acquisition. ................................................................................................................67
CHAPITRE 3 : INFLUENCE DES FACTEURS TECHNIQUES ...................................83
Article 1 : Réduction de dose dans l’exploration du rachis lombaire grâce au scanner 320-
détecteurs : étude initiale. .................................................................................................89
Article 2 : Amélioration de la qualité d’image scanographique en utilisant les
reconstructions itératives Adaptive Iterative Dose Reduction avec une acquisition wide-
volume sur un scanner 320-détecteurs...............................................................................99
Article 3 : Réduction de la dose des scanners abdominopelviens grâce aux reconstructions
itératives AIDR 3D. ........................................................................................................109
Article 4 : Scanner basse dose avec la modulation automatique du milliampérage, les
reconstructions Adaptive Statistical Iterative Reduction et un faible kilovoltage pour le
diagnostic des coliques néphrétiques : impact de l’indice de masse corporelle.................119
CHAPITRE 4 : APPLICATIONS CLINIQUES .............................................................135
Article 1 : Scanner basse dose pour la recherche d’une colique néphrétique : comment faire
en pratique clinique ?......................................................................................................141
Article 2 : Optimisation et réduction de la dose en scanner ostéo-articulaire....................151
Article 3 : Scanner ostéo-articulaire à large système de détection : principes, techniques et
applications en pratique clinique et en recherche.............................................................171
CONCLUSION GENERALE ET PERSPECTIVES ......................................................187
REFERENCES BIBLIOGRAPHIQUES ........................................................................193
RESUME ..........................................................................................................................204
ABSTRACT......................................................................................................................204

1
2
 ABREVIATIONS

AIDR Adaptive Iterative Dose Reduction

ALARA As Low As Reasonably Achievable (Aussi bas que raisonnablement possible)
ASIR Adaptive Statistical Iterative Reconstruction
CHRU Centre Hospitalier Régional Universitaire
FBP Filtered Back Projection (Rétroprojection Filtrée)
HIA Hôpital d’Instruction des Armées
IADI Imagerie Adaptative Diagnostique et Interventionnelle
IMC Indice de Masse Corporelle
IRM Imagerie par Résonance Magnétique
IRSN Institut de Radioprotection et de Sûreté Nucléaire
kV kilovoltage
mA milliampérage
mAs milliampère x secondes
mSv milliSievert
PDL Produit Dose Longueur
RLSS Relation Linéaire Sans Seuil
RCB Rapport Contraste sur Bruit
RSB Rapport Signal sur Bruit
T10 Dixième vertèbre thoracique

3
4
 TABLE DES FIGURES

Figure 1: Estimation du nombre de scanners réalisés annuellement aux Etats-Unis. ____________________ 21

Figure 2 : Fréquence des actes d’imagerie utilisant des rayonnements ionisants._______________________ 22

Figure 3 : Graphique représentant la relation entre la dose et les risques de cancer radio-induit.__________ 23

Figure 4 : Excès de risque relatif de la mortalité par cancer solide chez les survivants des bombes atomiques. 24

Figure 5 : Exemple montrant la qualité d’image d’une pomme au centre du scanner et à sa périphérie. _____ 28

Figure 6 : Bouclier de protection mammaire en bismuth chez une jeune femme.________________________ 29

Figure 7 : Bouclier de protection en bismuth pour la thyroïde. _____________________________________ 29

Figure 8 : Comparaison d’un scanner séquentiel et d’un scanner hélicoïdal. __________________________ 31

Figure 9 : Mise en évidence du phénomène d’overranging à partir d’un papier radiochromique. __________ 31

Figure 10 : Exemple de l’influence du milliampérage à partir de trois acquisitions d’un fantôme d’eau. ____ 32

Figure 11 : Représentation graphique de la modulation automatique du milliampérage. _________________ 33

Figure 12 : Exemple de l’influence du kilovoltage à partir de quatre acquisitions d’un fantôme d’eau.______ 34

Figure 13 : Exemple de l’influence du kilovoltage en cas de présence de produit de contraste iodé. ________ 34

Figure 14 : Scanner 320-détecteurs (Aquilion One®, Toshiba). ____________________________________ 36

Figure 15 : Scanner lombaire sans et avec les reconstructions itératives AIDR. ________________________ 37

Figure 16 : Scanner Optima CT660® installé à l’HIA Legouest.____________________________________ 37

Figure 17 : Fantôme Catphan® 500. _________________________________________________________ 38

Figure 18 : Dimension et pourcentage de contraste des tiges du module CTP515 du fantôme Catphan® 500. 38

Figure 19 : Coupe axiale du module CTP528 du fantôme Catphan® 500. ____________________________ 39

5
6
 TABLE DES TABLEAUX

Tableau 1 : Principaux facteurs comportementaux. ______________________________________________ 25

Tableau 2 : Paramètres techniques permettant d’optimiser la dose et la qualité d’image du scanner. _______ 30

7
8
INTRODUCTION GENERALE

9
10
 Depuis son introduction dans les années 1970, le scanner est devenu une technique
d’imagerie médicale incontournable notamment compte-tenu de son excellente performance
pour le diagnostic de nombreuses pathologies. Toutefois, le scanner est un examen d’imagerie
irradiant, son principe étant basé sur la reconstruction d’image à partir d’un faisceau de rayons
X. Même si les doses délivrées en scanographie sont faibles, de l’ordre du milliSievert (mSv),
compte-tenu des risques potentiels de cancer radio-induit lié aux faibles doses de rayons X, la
réduction de la dose d’irradiation au scanner est primordiale.

Cette thèse s’inscrit dans cette démarche d’optimisation et de réduction de la dose

d’irradiation au scanner. Elle est le fruit de plusieurs années de travail au sein du service
d’imagerie Guilloz du Centre Hospitalier Régional Universitaire (CHRU) de Nancy et du
service d’imagerie de l’Hôpital d’Instruction des Armées (HIA) Legouest de Metz, en
collaboration avec le laboratoire d’imagerie IADI (Imagerie Adaptative Diagnostique et
Interventionnelle) du CHRU de Nancy.

Le point de départ de ce travail a été l’installation en 2008 dans le service d’imagerie

Guilloz d’un nouveau type de scanner à large système de détection : le scanner 320-détecteurs
(scanner Aquilion One®, Toshiba Medical Systems, Otawara, Japon). L’originalité de ce
scanner était d’être composé d’un large système de détection de 16 cm dans l’axe z grâce à
320 rangées de détecteurs de 0,5 mm. Avec ce nouveau type de scanner, il devenait possible
de faire une acquisition d’un volume entier de 16 cm de longueur dans l’axe z en une seule
rotation du tube. Cette évolution technologique majeure a permis de pouvoir couvrir en une
rotation du tube des organes tels que le cœur, le cerveau ou encore des articulations
périphériques comme le poignet ou la cheville. Avec un temps de rotation minimal de 350 ms
et des techniques de reconstruction partielle des données, ce scanner bénéficie d’une
excellente résolution temporelle permettant une acquisition de 16 cm de données en moins de
175 ms. Grâce à l’introduction de ce nouveau type de scanner à large système de détection, il
était devenu possible de réaliser en pratique clinique courante des acquisitions de perfusion ou
encore des acquisitions dynamiques des articulations. Toutefois, pour ce type d’examen, de
nombreuses phases d’acquisitions doivent être réalisées, ce qui est à l’origine d’une
augmentation proportionnelle de la dose. Dans ce contexte, l’optimisation de la dose
d’irradiation du patient devient primordiale afin de pouvoir utiliser en pratique clinique
courante ce type de protocole.

11
 En 2010, dans l’optique de réduire la dose d’irradiation au scanner, une nouvelle
évolution technologique majeure est apparue sur le scanner 320-détecteurs : les
reconstructions itératives AIDR (Adaptive Iterative Dose Reduction). Ce type de
reconstruction a permis de réduire le bruit de l’image par rapport aux reconstructions standard
en rétroprojection filtrée (Filtered Back Projection - FBP). Ainsi, à qualité d’image
équivalente, l’implantation de ces reconstructions itératives était à l’origine d’une réduction
significative de la dose d’irradiation au scanner. En 2011, des reconstructions itératives ASIR
(Adaptive Statistical Iterative Reconstruction) ont également été disponibles sur un nouveau
scanner 64-détecteurs (Optima CT660®, GE Healthcare) installé dans le service d’imagerie
de l’HIA Legouest. Cette version des reconstructions itératives ASIR nous a notamment
permis de mettre au point un protocole de scanner basse dose pour le bilan des coliques
néphrétiques.

L’objectif principal de cette thèse est d’étudier les différentes manières d’optimiser et
de réduire la dose d’irradiation au scanner, tout en conservant une excellente performance
diagnostique. Pour cela, nous avons étudié sur fantôme et sur patient différents facteurs
techniques et comportementaux intervenant dans cette démarche d’optimisation et de
réduction de la dose au scanner. Nous nous sommes aussi intéressés à l’optimisation de
protocoles de scanner dans des applications cliniques particulières comme le scanner basse
dose réalisé pour le bilan d’une colique néphrétique et dans de nouvelles applications
cliniques avancées comme le scanner dynamique 4D des articulations ou le scanner de
perfusion tumorale.

Les objectifs de ce travail sont :

- L’évaluation de l’impact des facteurs comportementaux dans une démarche

d’optimisation et de réduction de la dose d’irradiation au scanner.

- L’évaluation sur fantôme et sur patient de plusieurs algorithmes de reconstruction

itérative (AIDR et AIDR 3D de Toshiba et ASIR de General Electrics) afin d’évaluer
leur performance et d’analyser leur impact en pratique clinique courante sur certains
types de pathologies comme la recherche d’une colique néphrétique.

- L’évaluation de la mise en œuvre en pratique clinique courante des techniques de

réduction et d’optimisation de la dose d’irradiation à la fois en imagerie
abdominopelvienne dans un contexte de recherche de colique néphrétique et en

12
 imagerie ostéo-articulaire, en particulier pour les nouvelles applications cliniques
comme le scanner dynamique des articulations, le scanner de perfusion ou encore le
scanner double-énergie.

Ce manuscrit est composé de quatre chapitres incluant l’ensemble des articles publiés ou
en cours d’évaluation pour publication sur la thématique de la réduction de la dose
d’irradiation au scanner. Le premier chapitre abordera des rappels concernant les risques
potentiels de cancer radio-induit lié aux faibles doses de rayons X et les différentes modalités
de réduction de la dose d’irradiation au scanner. Le deuxième chapitre s’intéressera à
l’influence des facteurs comportementaux sur la dose d’irradiation au scanner, le troisième à
l’influence des facteurs techniques et le quatrième et dernier chapitre portera sur des exemples
d’applications cliniques.

Le deuxième chapitre, portant sur l’influence des facteurs comportementaux, sera

composé de trois articles :

• Evaluation de la connaissance des prescripteurs de scanner en matière de

radioprotection des patients. Cet article a été publié dans le « Journal de
Radiologie » ;

• Evaluation de l’intérêt de l’acquisition abdominopelvienne sans injection lors

de la réalisation d’un scanner corps entier chez un patient suspect de
polytraumatisme. Cet article a été publié dans le journal « Diagnostic and
Interventional Imaging » ;

• Optimisation de la longueur d’acquisition des scanners réalisés pour colique

néphrétique : proposition d’une nouvelle méthode pour le placement de la
limite supérieure de l’acquisition. Cet article est en cours d’évaluation pour
publication dans le journal « British Journal of Radiology ».

Le troisième chapitre, portant sur l’influence des facteurs techniques, sera composé de
quatre articles :

• Réduction de dose dans l’exploration du rachis lombaire grâce au scanner 320-

détecteurs : étude initiale. Cet article a été publié dans le « Journal de
Radiologie » ;

13
 • Amélioration de la qualité d’image scanographique en utilisant les
reconstructions itératives Adaptive Iterative Dose Reduction avec une
acquisition wide-volume sur un scanner 320-détecteurs. Cet article a été publié
dans le journal « European Radiology » ;

• Réduction de la dose des scanners abdominopelviens grâce aux reconstructions

itératives AIDR 3D. Cet article a été publié dans le journal « Diagnostic and
Interventional Imaging » ;

• Scanner basse dose avec la modulation automatique du milliampérage, les

reconstructions Adaptive Statistical Iterative Reduction et un faible kilovoltage
pour le diagnostic des coliques néphrétiques : impact de l’indice de masse
corporelle. Cet article a été publié dans le journal « American Journal of
Roentgenology ».

Le quatrième et dernier chapitre, portant sur des exemples d’applications cliniques,

sera composé de trois articles :

• Scanner basse dose pour la recherche de colique néphrétique : comment faire

en pratique clinique ? Cet article a été publié dans le journal « Diagnostic and
Interventional Imaging » ;

• Optimisation et réduction de la dose en scanner ostéo-articulaire. Cet article a

été publié dans le journal « Diagnostic and Interventional Imaging » ;

• Scanner ostéo-articulaire à large système de détection : principes, techniques et

applications en pratique clinique et en recherche. Cet article a été publié dans
le journal « European Journal of Radiology ».

14
 LISTE DES PUBLICATIONS ET COMMUNICATIONS

PUBLICATIONS :

• Articles à comité de lecture international :

Gervaise A, Osemont B, Lecocq S, Micard E, Noel A, Felblinger J, Blum A. CT image

quality improvement using adaptive iterative dose reduction with wide-volume acquisition on
320-detector CT. Eur Radiol 2012; 22: 295-301.

Gervaise A, Teixeira P, Villani N, Lecocq S, Louis M, Blum A. Dose optimization and

reduction in musculoskeletal CT. Diagn Interv Imaging 2013; 94: 371-88.

Naulet P, Wassel J, Gervaise A, Blum A. Evaluation of the value of abdominopelvic

acquisition without contrast injection when performing a whole body CT scan in a patient
who may have multiple trauma. Diagn Interv Imaging 2013; 94: 410-7.

Gervaise A, Osemont B, Louis M, Lecocq S, Teixeira P, Blum A. Standard dose versus low-
dose abdominal and pelvic CT: comparison between filtered back projection versus adaptive
iterative dose reduction 3D. Diagn Interv Imaging 2014; 95: 47-53.

Gervaise A, Naulet P, Beuret F, Henry C, Pernin M, Portron Y, Lapierre-Combes. Low-dose

CT with automatic tube current modulation, adaptive statistical iterative reconstruction, and
low tube voltage for the diagnosis of renal colic. Am J Roentgenol 2014; 202: 553-60.

Teixeira P, Gervaise A, Louis M, Lecocq S, Raymond A, Aptel S, Blum A. Musculoskeletal

Wide detector CT : Principles, Techniques and Applications in Clinical Practice and Research.
Eur J Radiol 2015; 84: 892-900.

Blum A, Gervaise A, Teixeira P. Iterative reconstruction: Why, how and when? Diagn Interv
Imaging 2015; 96: 421-2.

Gervaise A, Gervaise-Henry C, Pernin M, Naulet P, Junca-Laplace C, Lapierre-Combes M.

Low dose CT for renal colic: how to do in clinical practice? Diagn Interv Imaging 2016; 97:
393-400.

Gervaise A, Teixeira P, Hossu G, Blum A, Lapierre-Combes C. Optimizing z-axis coverage

of abdominal CT scans of the urinary tract: a proposed alternative proximal landmark for
acquisition planning. En cours de soumission à British Journal of Radiology.

• Articles à comité de lecture national :

Gervaise A, Louis M, Batch T, Loeuille D, Noel A, Guillemin F, Blum A. Réduction de dose

dans l’exploration du rachis lombaire grâce au scanner 320-détecteurs : étude initiale. J
Radiol 2010; 91: 779-85.

Gervaise A, Esperabe-Vignau F, Naulet P, Pernin M, Portron Y, Lapierre-Combes M.

Evaluation des connaissances des prescripteurs de scanner en matière de radioprotection des
patients. J Radiol 2011; 92: 681-7.

15
COMMUNICATIONS :

• Communications affichées :

Gervaise A, Batch T, Loeuille D, Valckenaere I, Noël A, Guillemin F, Blum A. Comparaison

de la dose délivrée par un scanner 320-détecteurs en mode volumique versus hélicoïdal. 57ème
Journées Françaises de Radiologie, Paris, Octobre 2009.

Gervaise A, Esperabé-Vignau F, Pernin M, Naulet P, Portron Y, Lapierre-Combes M.

Evaluation des connaissances des prescripteurs en matière de radioprotection des patients.
59ème Journées Françaises de Radiologie, Paris, Octobre 2011.

Gervaise A, Micard E, Noel A, Felblinger J, Blum A. Evaluation des reconstructions

itératives AIDR sur le scanner 320-détecteurs : étude sur fantôme. 59ème Journées Françaises
de Radiologie, Paris, Octobre 2011.

Gervaise A, Osemont B, Lecocq S, Micard E, Noel A, Felblinger J, Blum A. CT image

quality improvement using adaptive iterative dose reduction with wide-volume acquisition on
320-detector CT. European Congress of Radiology, Vienna, Austria, 1-5 march 2012.

Villani N, Rovea F, Rossignon B, Gervaise A, Régent D. Comment améliorer le compromis

qualité image versus dose en radiologie diagnostique ? 51ème Journées Scientifiques de la
Société Française de Physique Médicale, Strasbourg, juin 2012.

Gervaise A, Osemont B, Louis M, Lecocq S, Teixeira P, Blum A. Réduction de la dose des

scanners abdominopelviens grâce aux reconstructions itératives AIDR 3D. 60ème Journées
Françaises de Radiologie, Paris, Octobre 2012.

Gervaise A, Teixeira P, Villani N, Lecocq S, Louis M, Blum A. Optimisation et réduction de

la dose en scanner ostéo-articulaire. 60ème Journées Françaises de Radiologie, Paris, Octobre
2012.

Naulet P, Wassel J, Gervaise A, Blum A. Evaluation de l’intérêt de l’acquisition

abdominopelvienne sans injection lors de la réalisation d’un scanner corps entier chez un
patient suspect de polytraumatisme. Journée des Assistants et des Internes, Ecole du Val de
Grace, Paris, 11 octobre 2012.

Gervaise A, Teixeira P, Villani N, Lecocq S, Louis M, Blum A. Musculoskeletal CT : How to

reduce the dose? RSNA 2012, Chicago, USA, 25-30 november 2012. Certificate of Merit.

Gervaise A, Osemont B, Louis M, Lecocq S, Teixeira P, Blum A. Full-dose versus low-dose

abdominal CT: comparison between Filtered Back Projection versus Adaptive Iterative Dose
Reduction 3D. European Congress of Radiology 2013, Vienna, Austria, 1-5 march 2013.

Gervaise A, Naulet P, Beuret F, Pernin M, Porton Y, Lapierre-Combes, Henry C. Evaluation

d’un protocole de scanner basse dose pour recherche d’une colique néphrétique. 61ème
Journées Françaises de Radiologie, Paris, 10 octobre 2013.

16
Beuret F, Naulet P, Lapierre-Combes M, Pernin M, Portron P, Gervaise A. Low-dose CT with
automatic tube current modulation, adaptive statistical iterative reconstruction, and low tube
voltage for the diagnosis of renal colic: impact of body mass index. RSNA 2013, Chicago,
USA, 1-6 December 2013.

Gervaise A, Beuret F, Naulet P, Henry C, Pernin M, Portron P, Lapierre-Combes M. Low-

dose CT with automatic tube current modulation, adaptive statistical iterative reconstruction,
and low tube voltage for the diagnosis of renal colic. European Congress of Radiology 2014,
Vienna, Austria, march 2014.

Gervaise A, Gervaise-Henry C, Pernin M, Naulet P, Junca-Laplace C, Lapierre-Combes M.

Low dose CT for renal colic: how to do it? European Congress of Radiology 2016, Vienna,
Austria, 4-8 march 2016.

Gervaise A, Teixeira P, Hossu G, Blum A, Lapierre-Combes M. Optimizing z-axis coverage

of abdominal CT scans of the urinary tract: a proposed alternative proximal landmark for
acquisition planning. European Congress of Radiology 2016, Vienna, Austria, 4-8 march
2016.

• Communications orales :

Gervaise A, Batch T, Loeuille D, Valckenaere I, Noël A, Guillemin F, Blum A. Comparaison

de la dose délivrée par un scanner 320-détecteurs en mode volumique versus hélicoïdal. 57ème
Journées Françaises de Radiologie, Paris, 17 octobre 2009.

Gervaise A, Pernin M, Naulet P, Junca-Laplace C, Darbois H, Lapierre-Combes M, Esperabe-

Vignau F. Evaluation des connaissances des praticiens de l’HIA Legouest en matière de
radioprotection des patients. 7ème Journée de Médecine d’Armée de l’Est, Metz, 21 octobre
2010.

Gervaise A. Reconstructions itératives : principes et applications dans la réduction de la dose

en scanographie. Séance de l’Association pour la Formation Permanente en Radiologie
(AFPR) : les nouveautés en radiologie ostéo-articulaire. Nancy, 24 février 2011.

Gervaise A. Impact des algorithmes itératifs de reconstruction sur l'interprétation des images
en scanographie. 50ème Journées Scientifiques de la Société Française de Physique Médicale,
Nantes, 9 juin 2011.

Gervaise A, Esperabe-Vignau F, Pernin M, Naulet P, Portron Y, Lapierre-Combes M.

Evaluation des connaissances des prescripteurs de scanner en matière de radioprotection des
patients. 59ème Journées Françaises de Radiologie, Paris, 21 octobre 2011.

Pernin M, Lapierre-Combes M, Naulet P, Junca-laplace C, Portron Y, Gervaise A. Evaluation

des connaissances des prescripteurs de scanner en matière de radioprotection des patients.
Journée des Assistants et des Internes, Ecole du Val de Grace, Paris, octobre 2011.

Bisconte S, Manen O, Dubourdieu D, Gervaise A, Hornez AP, Oliviez JF, Deroche J, Brunetti
G, Martel V, Perrier E, Généro M. Réflexions sur l'irradiation médicale en expertise
aéronautique. Réunion de la SOFRAMAS, Paris, 19 janvier 2012.

17
Blum A, Gervaise A, Teixeira P. AIDR 3D: a promising dose reduction tool in clinical
applications. European Congress of Radiology, Vienna, Austria, 4 march 2012.

Gervaise A, Osemont B, Lecocq S, Micard E, Noel A, Felblinger J, Blum A. CT image

quality improvement using adaptive iterative dose reduction with wide-volume acquisition on
320-detector CT. European Congress of Radiology, Vienna, Austria, 5 march 2012.

Gervaise A. Optimisation et réduction de la dose des scanners en pathologie ostéo-articulaire.

Séance de l'Association pour la Formation Permanente en Radiologie (AFPR) : les nouveautés
en radiologie ostéo-articulaire, Nancy, 6 mars 2012.

Blum A, Teixeira P, Villani N, Lecocq-Teixeira S, Louis M, Gervaise A. How to reduce the

CT-scan radiation dose in MSK disorders? 41ème Congrès de radiologie brésilienne, Brasilia,
Brésil, 6-8 septembre 2012.

Beuret F, Gervaise A, Naulet P, Pernin M, Portron Y, Lapierre-Combes M. Evaluation d’un

scanner basse dose chez des patients adressés pour recherche de colique néphrétique. Journée
de printemps de la SFR Lorraine, Metz, 13 avril 2013. Prix de la meilleure communication
libre.

Naulet P, Junca-Laplace C, Pernin M, Portron Y, Lapierre-Combes M, Gervaise A. Impact du

remplacement d’un scanner par un de nouvelle génération sur la qualité d’image et
l’exposition des patients aux rayonnements ionisants. Journée de printemps de la SFR
Lorraine, Metz, 13 avril 2013.

Gervaise A, Naulet P, Beuret F, Pernin M, Portron Y, Lapierre-Combes M. Intérêt du scanner

basse dose dans l’exploration des coliques néphrétiques de l’adulte. 10ème Journée de
Médecine d’Armée de l’Est, Metz, 24 octobre 2013.

Gervaise A, Naulet P, Beuret F, Henry C, Pernin M, Portron P, Lapierre-Combes C. Low-

dose CT with automatic tube current modulation, adaptive statistical iterative reconstruction,
and low tube voltage for the diagnosis of renal colic: impact of body mass index. RSNA,
Chicago, USA, 1st December 2013.

Gervaise A. Scanner SubmSv : état des lieux et perspectives. VII Symposium scanner
volumique, Nancy, 26 janvier 2016.

18
CHAPITRE 1 : CONTEXTE ET PROBLEMATIQUE

19
20
 1- Contexte :

Développé au début des années 1970, le scanner s’est imposé de nos jours comme une
technique d’imagerie médicale incontournable. Grâce à des développements technologiques
considérables ces dernières années (scanner hélicoïdal en 1989, scanner multi-détecteurs en
1998, scanner bi-tube en 2006, scanner à large système de détection en 2008) et à sa grande
disponibilité, le scanner est devenu l’examen de choix pour l’exploration de nombreuses
pathologies traumatiques, osseuses, pulmonaires, cardio-vasculaires ou encore néoplasiques.
Il permet d’en faire le diagnostic, d’en suivre l’évolution et peut même en permettre le
traitement à travers l’émergence de la radiologie interventionnelle.

Compte tenu de ses performances, le nombre de scanners réalisés chaque année est en
constante progression. Aux Etats-Unis, ce sont plus de 70 millions de scanners réalisés chaque
année [1] (Figure 1). En France, l’Institut de Radioprotection et de Sûreté Nucléaire (IRSN)
rapporte une augmentation de 26 % des actes de scanographie entre 2002 et 2007 [2] et une
augmentation de 12 % entre 2007 et 2012 [3] avec plus de 8 millions de scanners réalisés
chaque année.

Figure 1: Estimation du nombre de scanners réalisés annuellement aux Etats-Unis (d’après Brenner DJ et
al. 2007 [1]).

21
 Toutefois, le scanner est une technique d’imagerie irradiante. Son principe est basé sur
la reconstruction d’une image à partir du calcul de l’atténuation de multiples projections d’un
faisceau de rayons X autour du patient. Même si la dose de rayons X délivrée au décours d’un
scanner reste faible, de l’ordre du mSv, le scanner est responsable en France de 71 % de
l’irradiation due aux rayonnements ionisants d’origine médicale alors que les actes de
scanographie ne représentent que 10,1 % de l’ensemble des actes de radiologie irradiants
(Figure 2) [3].

Figure 2 : Fréquence des actes d’imagerie utilisant des rayonnements ionisants et répartition par modalité
d’imagerie de la dose efficace collective en 2012 (d’après le rapport IRSN 2014 [3]).

Or, même si le lien entre l’exposition à de faibles doses de rayons X et l’augmentation

du risque de cancer radio-induit est fortement controversé, il a été établi par plusieurs grandes
institutions et par de nombreux rapports ou publications internationales [4-6]. En particulier,
la conférence de consensus BEIR VII (Biologic Effects of Ionizing Radiation) qui s’est tenue
en 2006 et qui s’appuyait sur les estimations les plus récentes et complètes en matière de
risque de cancer suite à l’exposition à de faibles doses de rayonnements ionisants soutient un
modèle de relation linéaire sans seuil (RLSS) [4]. Cela signifie que le risque de cancer causé
par de faibles doses de rayonnements ionisants augmente de façon linéaire et ne connait pas
de seuil (Figure 3). De ce fait, la plus petite dose a le potentiel de provoquer une légère
augmentation du risque de cancer radio-induit chez l’homme. En reprenant ce modèle de
RLSS des effets des rayons X à faibles doses, des publications ont ainsi estimé le risque de
cancer radio-induit lié à la réalisation d’un seul scanner abdominopelvien (avec une dose
d’environ 10 mSv) de l’ordre de 1/1000 chez une jeune femme [7].

22
Figure 3 : Graphique représentant la relation entre la dose d’irradiation (Radiation Dose) et les risques de
cancer radio-induit (Risk). La ligne continue représente les données connues pour les doses supérieures à
200 mSv, avec une relation linéaire et proportionnelle entre le niveau de dose d’irradiation et le risque de
cancer radio-induit. Pour les faibles doses (< 200 mSv), la ligne en pointillé correspond à l’extrapolation
des données suivant un modèle de régression linéaire sans seuil (source internet :
[Link]

Ce risque important reste toutefois un risque maximal théorique. D’autres auteurs

estiment pour leur part soit qu’un tel risque n’existe pas, soit qu’il est largement surestimé [8-
10]. Enfin, des résultats expérimentaux récents tendent à montrer que suite à une exposition à
de faibles doses de rayons X, les effets biologiques ne suivent pas une relation linéaire mais
entraineraient des effets positifs selon une réponse adaptative. Cet effet « Hormesis » serait lié
à la stimulation des mécanismes de réparation tissulaire secondaire à l’irradiation des cellules
par une faible dose de rayons X et apporterait une protection de la cellule vis-à-vis des
rayonnements ionisants [10] (Figure 4).

23
 Figure 4 : Excès de risque relatif de la mortalité par cancer solide chez les survivants des bombes
atomiques : notez l’inversion du risque relatif pour les doses les plus faibles traduisant l’effet « Hormesis »
(d’après Doss M. 2013 [10]).

L’absence de preuves scientifiques formelles allant dans un sens ou dans l’autre impose de
suivre le principe de précaution et aboutit à la nécessité de réduire au maximum les doses
d’irradiation délivrées aux patients. Ce principe de précaution ALARA (« As Low As
Reasonably Achievable ») a d’ailleurs été repris par l’Union Européenne dans la directive
Euratom 97/43 [11] puis par la directive Euratom 2013/59 [12]. Cette dernière précise que les
examens d’imagerie irradiants doivent faire l’objet d’une justification et d’une optimisation
constante afin de réduire les doses individuelles et collectives dues aux expositions médicales.
Toutefois, même si la réduction des doses délivrées au scanner est devenue primordiale, elle
ne doit pas se faire aux dépens de la performance diagnostique des examens, le but étant
d’obtenir l’information diagnostique avec la dose de rayons X la plus faible. C’est ainsi que
ces dernières années, de nombreux efforts ont été réalisés pour réduire les doses délivrées au
cours des scanners.

24
 2- Problématique : comment optimiser et réduire la dose d’irradiation au scanner ?

Les modalités d’optimisation et de réduction de la dose d’irradiation au scanner sont

nombreuses. Elles correspondent soit à des facteurs comportementaux, soit à des facteurs
techniques.

A- Les facteurs comportementaux :

Les facteurs comportementaux correspondent à l’ensemble des mesures qui permettent de

réduire la dose d’irradiation et dont la mise en œuvre est indépendante des facteurs techniques
et du matériel scanographique disponible par le radiologue (Tableau I).

Tableau 1 : Principaux facteurs comportementaux à prendre en compte dans une démarche

d’optimisation et de réduction de la dose d’irradiation au scanner.

Principaux facteurs comportementaux

- Education et sensibilisation des équipes médicales et paramédicales

- Justification de l’indication des scanners
- Substitution des scanners par une technique d’imagerie non irradiante
- Réduction du nombre de phases d’acquisition
- Limitation de la couverture d’acquisition
- Positionnement optimal du patient
- Utilisation de bouclier de protection en bismuth

L’intérêt principal de ces facteurs comportementaux est qu’ils sont indépendants du type
de matériel disponible par le radiologue et donc qu’ils sont utilisables pour n’importe quel
type de scanner. Il s’agit aussi de méthodes de réduction de la dose qui sont peu coûteuses,
simples et rapides à mettre en œuvre. Nous proposons de faire un bref rappel sur l’intérêt de
chacun de ces facteurs comportementaux dans une démarche d’optimisation et de réduction
de la dose d’irradiation au scanner.

25
 a- Education et sensibilisation :

L’éducation et la sensibilisation des équipes médicales et paramédicales sont

fondamentales [13]. Cette sensibilisation nécessite toutefois la connaissance des doses
délivrées par les radiologues et les manipulateurs en radiologie. Dans cette optique,
l’affichage du PDL (Produit Dose Longueur) à la console du scanner avant la réalisation de
l’acquisition est indispensable et est actuellement systématiquement disponible pour
l’ensemble des constructeurs. La sensibilisation des radiologues et des manipulateurs en
radiologie est également de plus en plus assurée par le biais de logiciel de recueil et d’analyse
des doses délivrées. Outre la connaissance des doses délivrées pour chaque scanner, ces
logiciels permettent le suivi dosimétrique par patient et permettent de détecter pour certains
des cumuls de dose parfois importants. Ces logiciels comportent également des alertes
dosimétriques qui incitent à optimiser les protocoles et apportent une sécurité dosimétrique
vis-à-vis des patients. Ils permettent enfin de suivre la réduction globale des doses au cours
d’une démarche d’optimisation des protocoles scanographiques [14]. De manière plus globale,
des registres de doses nationaux ou internationaux sont également en train de voir le jour,
comme par exemple le CT Dose Index Registry [15]. Ce registre mis en service par l’initiative
de l’American College of Radiology a pour but de récupérer les doses scanographiques de
nombreux services de radiologie américains et étrangers dans l’optique de comparer les doses
et d’harmoniser les pratiques.

Enfin, l’éducation des médecins et des manipulateurs en radiologie permet de mettre en

avant les situations pour lesquelles la réduction de dose est particulièrement importante.
Notons, par exemple, que l’âge est un facteur primordial puisque le risque potentiel de cancer
radio-induit lié aux faibles doses de rayons X décroit avec l’âge [16]. Une vigilance
particulière est donc la règle chez les sujets jeunes. De même, la localisation anatomique du
scanner est importante à prendre en compte. La dose efficace d’une acquisition à distance des
organes radiosensibles, comme c’est le cas pour les articulations périphériques, sera
négligeable (avec des doses efficaces parfois inférieures à celle d’une radiographie
thoracique !) au contraire d’une acquisition thoraco-abdominopelvienne.

26
 b- Justification et substitution :

La justification et la substitution d’un scanner par une technique d’imagerie non irradiante
comme l’échographie ou l’Imagerie par Résonance Magnétique (IRM) sont aussi deux
éléments importants dans une démarche de réduction de la dose au scanner : « le scanner qui
irradie le moins est celui qui n’est pas réalisé ». Par exemple, Oikarinen H et al. [17] ont
montré dans leur étude portant sur 30 scanners lombaires réalisés chez des patients de moins
de 35 ans, que seulement 7 (23 %) étaient indiqués. Sur les 23 scanners lombaires non
indiqués, 20 auraient pu bénéficier d’une IRM tandis que pour 3 patients, il n’y avait aucune
indication d’imagerie.

c- Réduction du nombre de phases d’acquisition :

La limitation du nombre de phases d’acquisition d’un scanner permet aussi de réduire la

dose. En effet, la dose globale de l’examen est calculée à partir de la somme des doses de
chaque acquisition. Par exemple, lors de la réalisation d’un scanner abdominopelvien, la
réalisation d’une unique série injectée au temps portal versus deux acquisitions identiques
sans et après injection au temps portal permet de réduire la dose de moitié. Tandis que
certaines pathologies nécessitent la réalisation d’un examen multiphasique avec une série sans
injection et deux ou trois acquisitions après injection de produit de contraste iodé à différents
temps (exploration des lésions hépatiques, caractérisation d’une lésion surrénalienne, bilan
d’une lésion tumorale rénale) [18-19], pour d’autres pathologies (embolie pulmonaire, poly-
traumatisme) des études ont montré l’absence d’amélioration de la performance diagnostique
liée à la série sans injection [20].

d- Limitation de la couverture d’acquisition :

La limitation de la couverture d’acquisition est un moyen simple, efficace et rapide pour

réduire la dose. En effet, même si la longueur d’acquisition n’affecte pas l’indice de dose
scanographique volumique, la dose totale de l’examen est directement liée à la longueur
d’acquisition et au risque de cancer radio-induit lié aux faibles doses de rayons X. La
couverture d’acquisition doit être centrée sur la zone d’intérêt, préalablement repérée par le ou
les topogramme(s) du scanner. Avec les scanners modernes, permettant une acquisition rapide

27
d’un grand volume, les couvertures d’acquisition des scanners ont tendance à augmenter [21].
Tandis qu’une longueur d’acquisition trop importante est à l’origine d’une augmentation de la
dose délivrée, à l’inverse, une réduction trop importante de la couverture d’acquisition peut
être à l’origine d’une baisse de la performance diagnostique par la non-visualisation d’une
structure pathologique située en dehors de la zone explorée [22]. Par exemple, pour la
recherche d’une embolie pulmonaire, une réduction de la couverture d’acquisition jusqu’à
47 % permet de conserver une bonne performance diagnostique pour la recherche d’une
embolie pulmonaire [23] mais une réduction trop importante de la longueur d’acquisition peut
aussi masquer des diagnostics différentiels [22].

e- Positionnement optimal du patient :

Un centrage précis de la zone anatomique à scanner au centre de l’anneau permet

d’obtenir une qualité d’image et une dose délivrée optimales [24]. La résolution spatiale est
effectivement meilleure au centre de l’anneau car plus d’interpolations des données y sont
réalisées par rapport à la périphérie [25] (Figure 5). Par ailleurs, un bon centrage est
particulièrement nécessaire lors de l’utilisation de la modulation automatique du
milliampérage car celle-ci considère le patient au centre de l’anneau. En cas de mauvais
centrage, la modulation automatique fait augmenter la dose de façon significative [26]. La
position du patient influe également sur la dose et la qualité d’image. L’épaisseur du volume
scanné doit être la plus fine possible pour limiter les artéfacts de durcissement du faisceau.

Figure 5 : Exemple montrant la qualité d’image d’une pomme positionnée au centre de l’anneau du
scanner (a) et à sa périphérie (b). Notez la dégradation de la qualité d’image et de la résolution spatiale
quand la pomme n’est pas centrée (b).

28
 f- Utilisation de bouclier de protection en bismuth

Des boucliers de protection en bismuth sont disponibles pour diminuer la dose

d’irradiation aux glandes mammaires lors de l’acquisition d’un scanner thoracique (Figure 6),
pour la thyroïde au cours d’un scanner cervical (Figure 7) et pour protéger les yeux lors de
l’acquisition d’un scanner cérébral. Par exemple, lors de l’acquisition d’un scanner cérébral
avec un bouclier de protection oculaire, Ciarmatori A et al. rapportent une réduction de la
dose au cristallin de l’ordre de 21 à 28 %, sans modification de la qualité d’image du scanner
cérébral [27].

Figure 6 : Bouclier de protection mammaire en bismuth chez une jeune femme. Notez les artéfacts de
durcissement du faisceau au niveau des seins et au contact du bouclier de protection traduisant une
absorption plus importante des rayons X par le bouclier et permettant ainsi de réduire la dose aux glandes
mammaires. Par contre, l’analyse du thorax ne montre pas d’altération de la qualité d’image.

Figure 7 : Bouclier de protection en bismuth pour la thyroïde (source internet :

[Link]

29
 B- Facteurs techniques :

Les facteurs techniques sont ceux liés au type et au modèle de scanner utilisé. Ces facteurs
sont nombreux. Certains ne sont pas modifiables directement par l’utilisateur comme par
exemple la géométrie du couple tube-détecteur, le type de détecteur, la filtration ou encore
l’utilisation d’un bouclier anti-hélice. D’autres paramètres sont accessibles et peuvent être
modifiés pour optimiser la dose et la qualité d’image, soit au moment de l’acquisition des
images, soit a posteriori. Le tableau II reprend les différents paramètres techniques accessibles
dans une démarche d’optimisation de la dose.

Tableau 2 : Paramètres techniques permettant d’optimiser la dose et la qualité d’image du scanner au

moment de l’acquisition ou a posteriori.

Paramètres accessibles au moment de

Paramètres accessibles après l’acquisition
l’acquisition

- Mode d’acquisition : séquentiel ou hélicoïdal

- Tension du tube radiogène (kilovoltage) - Epaisseur de coupe et l’intervalle inter-coupe
- Charge du tube radiogène (milliampérage) - Algorithme de reconstruction
- Nombre de détecteur - Filtre de reconstruction
- Pas de l’hélice (pitch)

a- Mode d’acquisition :

Le mode d’acquisition est l’un des facteurs techniques qui influence la dose et qui est
accessible au moment de l’acquisition. Historiquement, les premières acquisitions
scanographiques étaient réalisées en mode axial séquentiel (Figure 8a). Avec ce mode
d’acquisition, le couple tube-détecteur fait une rotation autour du patient puis la table avance
avant de refaire une rotation et ainsi de suite. Dans les années 2000, le développement de
l’acquisition hélicoïdale a permis de réduire considérablement le temps d’acquisition. En
mode hélicoïdal, l’acquisition est continue et la table avance en même temps que l’acquisition
est réalisée, sans marquer d’arrêt (Figure 8b).

30
 Figure 8 : Comparaison d’un scanner séquentiel (a) et d’un scanner hélicoïdal (b) (source internet :
[Link]

Avec le mode hélicoïdal, afin de s’assurer de l’entièreté des premières et dernières

coupes de l’acquisition, il est nécessaire de faire un tour supplémentaire à chaque extrémité de
la zone explorée. Cette exposition « pré et post-hélice » appelée overranging ou z
overscanning varie proportionnellement en fonction du pitch et du nombre de détecteurs. Le
pourcentage de dose liée à l’overranging est inversement proportionnel à la longueur
d’exploration [28]. De ce fait, avec les scanners modernes comportant fréquemment une
rangée de 64-détecteurs, l’overranging peut représenter une part non négligeable de
l’irradiation totale (de l’ordre de 10 % pour une acquisition abdominopelvienne et jusqu’à 20
ou 30 % pour une acquisition cardiaque) [29]. Plus récemment, les avancées technologiques
ont permis de mettre au point des scanners à large système de détection. Par exemple, au
CHRU de Nancy, le scanner 320-détecteurs a été installé en 2008 permettant de faire une
acquisition de 16 cm en une rotation. Il est donc possible de revenir au mode d’acquisition
séquentiel « wide-volume » et de réduire la dose par rapport à une acquisition hélicoïdale
classique en supprimant l’irradiation inutile liée à l’overranging (Figure 9).

Figure 9 : Mise en évidence du phénomène d’overranging à partir d’un papier radiochromique (en
présence de rayons X, la zone irradiée s’assombrit). Deux acquisitions de même longueur sont réalisées en
mode hélicoïdal (papier du haut) et volumique séquentiel (papier du bas). Pour la même longueur
d’acquisition, la zone irradiée est plus importante en mode hélicoïdal : cette exposition « pré et post-
hélice » correspond à l’overranging (doubles flèches).

31
 b- Milliampérage :

Lors de l’acquisition, le principal facteur accessible et modifiable par l’utilisateur est le

milliampérage (mA). Le mA correspond à l’intensité du courant du tube radiogène et
représente la quantité de photons produite au sein du faisceau de rayons X. Le mA est
proportionnel à la dose délivrée et inversement proportionnel au carré du bruit de l’image [30].
Une réduction du milliampérage par 2 permet donc de réduire la dose par 2 mais augmente le
bruit de l’image d’un facteur 1,4 (Figure 10).

Figure 10 : Exemple de l’influence du milliampérage à partir de trois acquisitions d’un fantôme d’eau
avec des valeurs de milliampérage décroissant (400, 200 et 100 mA) et en gardant les autres paramètres
d’acquisition constants. Notez la dégradation de la qualité d’image avec une augmentation du bruit quand
le milliampérage diminue.

L’optimisation du milliampérage doit aussi prendre en compte le morphotype du

patient. En effet, la quantité de photons X nécessaire pour garder une qualité de l’image
constante est variable en fonction de l’épaisseur du patient. Pour mieux adapter le mA au
morphotype des patients, des techniques de modulation automatique du milliampérage ont été
développé au début des années 2000 [31]. Grâce à ces techniques, le mA est automatiquement
adapté au morphotype du patient et à la zone anatomique à explorer et le bruit de l’image reste
constant sur l’ensemble des images de l’acquisition [32] (Figure 11).

32
 Figure 11 : Représentation graphique de la modulation automatique du milliampérage sur un scanner
thoracique. La modulation longitudinale du milliampérage est caractérisée par un changement du
milliampérage dans l’axe z en fonction des changements de l’atténuation du patient (A). Une modulation
angulaire permet aussi d’adapter le milliampérage au sein d’une même coupe du fait de l’asymétrie de
l’atténuation entre les régions antéropostérieures et latérales de la coupe (B) (d’après Singh S et al. 2011
[32]).

c- Kilovoltage :

Le kilovoltage (kV) correspond à l’énergie des photons produits au sein du faisceau de

rayons X. Il est généralement compris entre 70 et 140 kV. La baisse du kilovoltage est à
l’origine d’une réduction importante de la dose (par exemple, en gardant les autres paramètres
constants, la baisse du kilovoltage de 120 à 80 kV réduit la dose délivrée d’un facteur 2,2
[26]) mais est aussi à l’origine d’une augmentation du bruit [33] (Figure 12). Une réduction
trop importante du kV peut aussi être à l’origine d’artéfact de durcissement du faisceau. Par
contre, la réduction du kV améliore le contraste des images réalisées avec injection de produit
de contraste iodé car l’abaissement de l’énergie moyenne du faisceau favorise d’absorption
des photons X par effet photo-électrique et cela d’autant plus que l’énergie des rayons X est
proche du pic de fluorescence de l’iode (Figure 13).

33
Figure 12 : Exemple de l’influence du kilovoltage à partir de quatre acquisitions d’un fantôme d’eau avec
des valeurs de kV croissant (80, 100, 120 et 140 kV) et en gardant les autres paramètres d’acquisition
constants. En diminuant le kV de 120 à 80 kV la dose est réduite d’un facteur 2,2 mais le bruit augmente.

Figure 13 : Exemple de l’influence du kilovoltage en cas de présence de produit de contraste iodé.

Acquisitions d’un verre contenant 100 mL d’eau et 5 mL d’IOMERON 400® avec des valeurs de kV
croissante (80, 100, 120 et 140 kV) et en gardant les autres paramètres constants. Notez l’augmentation
progressive des valeurs d’atténuation quand le kV diminue.

d- Pitch ou pas de l’hélice :

Avec les scanners multi détecteurs actuels comportant les techniques modernes de
modulation de la dose, le changement du pitch (pas de l’hélice) ne modifie plus la dose car il
s’ensuit une adaptation automatique du milliampérage [34]. Un pitch élevé, de l’ordre de 1,5,
sera préféré pour réduire le temps d’acquisition et les artéfacts de mouvement (par exemple,
lors de l’exploration d’un patient polytraumatisé). Le pitch doit toutefois rester inférieur à 2
afin de garder une qualité optimale des reformations multi-planaires [34] et d’éviter
l’apparition d’artéfacts d’hélice [35]. A l’opposé, un petit pitch sera préféré pour réduire les
artéfacts métalliques en rapport avec les matériels d’ostéosynthèse [36].

34
 e- Epaisseur de coupe :

De manière générale, les acquisitions sont réalisées en coupes fines (0,5 à 1 mm) et sont
reconstruites en coupes plus épaisses (2 à 5 mm). Les coupes submillimétriques améliorent la
résolution spatiale, réduisent les effets de volume partiel et permettent la réalisation de
reconstruction dans un volume quasi-isotrope [37]. Par contre, à bruit constant, l’acquisition
en coupes fines est à l’origine d’une augmentation de l’irradiation [38]. En cas de réduction
excessive du milliampérage, l’acquisition en coupes fines engendre une augmentation
importante du bruit de l’image. Ainsi, tandis que l’acquisition se fait en coupes
submillimétriques, lors de l’interprétation des images, l’épaississement des coupes permet
d’augmenter le rapport signal sur bruit [37] et d’améliorer l’analyse des images [39-40].

f- Reconstructions des images :

La reconstruction des images est un facteur technique important qui est modifiable après
l’acquisition et qui influence la qualité des images et donc indirectement la dose.
Historiquement, les images des scanners étaient reconstruites à partir d’une rétroprojection
filtrée. Cette méthode avait pour avantage d’être simple, robuste et rapide. Par contre, elle
était à l’origine d’un bruit important de l’image, notamment en cas de réduction trop
importante de la dose [41]. Entre 2008 et 2010, grâce à l’importante augmentation de la
puissance informatique, les principaux constructeurs de scanner ont commercialisés des
nouveaux algorithmes de reconstruction itérative. Grâce à une meilleure utilisation des
données issues des projections, ces algorithmes ont permis de réduire le bruit des images
scanographiques [42]. A qualité d’image constante, ces algorithmes ont donc permis de
réduire la dose des scanners [43].

g- Filtre de reconstruction :

L’amélioration du rapport contraste sur bruit peut également se faire par l’utilisation de
filtres de réduction de bruit à partir de logiciels de post-traitement. L’application de ces filtres
se fait sur des images déjà reconstruites ce qui permet de les utiliser à partir de n’importe
quelle image scanner, y compris sur des reformations 3D.

35
 3- Environnement de travail :

Au cours de notre travail de thèse, nous avons principalement travaillé dans le service
d’imagerie Guilloz du CHRU de Nancy et dans le service d’imagerie médicale de l’HIA
Legouest de Metz.

Dans le service d’imagerie Guilloz, nous avons travaillé sur le scanner 320-détecteurs
(Figure 14). Ce scanner a été installé en 2008. Il s’agissait à l’époque d’une révolution car il
était le premier scanner à large système de détection installé en France.

Grâce à sa rangée de 320 détecteurs de 0,5 mm, ce scanner permet l’acquisition de 16 cm

de données dans l’axe z en une seule rotation du tube. Son temps de rotation minimal de 350
ms couplé à des techniques de reconstruction partielle des données permet ainsi d’avoir une
excellente résolution temporelle de l’ordre de 175 ms. La répétition de multiples volumes
d’acquisition a aussi permis de mettre au point des protocoles de perfusion tumorale ou
encore de scanner dynamique des articulations.

Figure 14 : Scanner 320-détecteurs (Aquilion One®, Toshiba). Ce scanner est un scanner à large système
de détection (16 cm) comprenant 320 rangées de détecteurs de 0,5 mm. Grâce à ce scanner il est possible
de faire l’acquisition en une seule rotation de 16 cm de données.

C’est aussi sur ce scanner qu’a été installé en 2010 pour la première fois au CHRU de
Nancy un algorithme de reconstruction itérative : les reconstructions AIDR. Grâce à ces

36
reconstructions itératives, il était possible de réduire le bruit des images par rapport aux
reconstructions standard en FBP (Figure 15).

Figure 15 : Scanner lombaire sans (a) et avec les reconstructions itératives AIDR (b). Notez la réduction
du bruit de l’image de 34 % grâce aux reconstructions itératives AIDR.

Dans le service d’imagerie de l’HIA Legouest, nous avons eu accès à un scanner 64-
détecteurs Optima CT660®. Ce scanner correspond à ce qui se fait classiquement de nos jours
dans les services d’imagerie avec un scanner de 64-détecteurs de 0,625 mm et avec un temps
de rotation minimal de 500 ms (Figure 16). En même temps que l’installation de ce scanner
fin 2011 a été installé les reconstructions itératives ASIR.

Figure 16 : Scanner Optima CT660® installé à l’HIA Legouest.

37
 Au cours de ce travail, nous avons aussi utilisé un fantôme Catphan® 500 (The
Phantom Laboratory, Salem, NY, USA) afin de faire des mesures de qualité d’image
objective (Figure 17).

Figure 17 : Fantôme Catphan® 500 (The Phantom Laboratory, Salem, NY, USA) (source internet :
[Link]

Ce fantôme est composé de plusieurs modules. Pour calculer le bruit de l’image, le

rapport signal sur bruit (RSB) et le rapport contraste sur bruit (RCB) nous avons utilisé le
module CTP515. Ce module à faible contraste contient des tiges cylindriques de différents
diamètres et trois différents niveaux de contraste (Figure 18).

Figure 18 : Dimension et pourcentage de contraste des tiges cylindriques présentes dans le module
CTP515 du fantôme Catphan® 500 (d’après Catphan® 500 and 600 Manual, Copyright © 2012).

38
 Pour calculer la résolution spatiale, nous avons utilisé le module 528. Ce module à
haute résolution contient deux structures à haute densité pour évaluer la résolution spatiale du
scanner (Figure 19). La première structure est un motif circulaire comprenant 21 éléments
dont la résolution varie de 1 jusqu’à 21 paires de ligne par cm. La seconde structure comprend
deux billes sphériques encastrées dans un matériau uniforme (flèche).

Figure 19 : Coupe axiale du module CTP528 du fantôme Catphan® 500 (d’après Catphan® 500 and 600
Manual, Copyright © 2012).

39
40
CHAPITRE 2 : INFLUENCE DES FACTEURS COMPORTEMENTAUX

Ce chapitre est composé de trois articles :

1- Gervaise A, Esperabe-Vignau F, Naulet P, Pernin M, Portron Y, Lapierre-Combes M.

Evaluation des connaissances des prescripteurs de scanner en matière de
radioprotection des patients. J Radiol 2011; 92: 681-87.

2- Naulet P, Wassel J, Gervaise A, Blum A. Evaluation of the value of abdominopelvic

acquisition without contrast injection when performing a whole body CT scan in a
patient who may have multiple trauma. Diagn Interv Imaging 2013; 94: 410-7.

3- Gervaise A, Teixeira P, Hossu G, Blum A, Lapierre-Combes C. Optimizing z-axis

coverage of abdominal CT scans of the urinary tract: a proposed alternative proximal
landmark for acquisition planning. En cours de soumission à British Journal of
Radiology.

41
42
 Les facteurs comportementaux jouent un rôle primordial dans une démarche
d’optimisation et de réduction de la dose d’irradiation au scanner. Nous proposons d’illustrer
l’intérêt de trois de ces facteurs au cours d’une démarche d’optimisation et de réduction de la
dose au scanner : la sensibilisation des médecins prescripteurs par l’évaluation de leurs
connaissances en matière de radioprotection des patients, la réduction du nombre de phases
d’acquisition et la limitation de la couverture d’acquisition des scanners.

Article 1 : Evaluation des connaissances des prescripteurs de scanner en matière de

radioprotection des patients.

La justification des examens d’imagerie irradiants est un des deux grands principes de la
radioprotection [12]. Il s’agit d’une responsabilité partagée entre le prescripteur et le
radiologue. Son application nécessite toutefois de connaître et de prendre en compte les
risques potentiels de cancer radio-induit lié aux faibles doses de rayons X. Pourtant, de
nombreuses études ont montré le défaut de connaissance des médecins prescripteurs en
matière de radioprotection des patients [44-45].

Le but de cette étude était d’évaluer les connaissances des praticiens prescripteurs de
scanner en matière de radioprotection des patients. Il s’agissait de la première étude portant
sur ce sujet qui a été conduite en France. Cette étude a été réalisée à partir d’un questionnaire
envoyé par courrier à l’ensemble des praticiens hospitaliers de l’HIA Legouest durant le mois
d’avril 2010. Le questionnaire était composé d’une première partie analysant les données
démographiques des praticiens et d’une deuxième partie comprenant huit questions et
abordant plusieurs thèmes testant les connaissances des praticiens en matière de
radioprotection des patients.

Les résultats montraient que 70 % des praticiens déclaraient prendre en compte les risques
liés aux rayons X lors de la prescription d’un scanner. Par contre, la connaissance des doses
délivrées lors de la réalisation d’un scanner abdominopelvien était mal maitrisée et les risques
potentiels de cancer radio-induit lié aux faibles doses de rayons X étaient largement sous-
estimés. Enfin, seulement 34 % des praticiens avaient bénéficié d’une formation à la
radioprotection des patients.

43
 Article 2 : Evaluation de l’intérêt de l’acquisition abdominopelvienne sans injection lors
de la réalisation d’un scanner corps entier chez un patient suspect de polytraumatisme.

La limitation du nombre de phases d’acquisition est un moyen simple pour réduire la dose
globale d’un scanner. Par exemple, lors de la réalisation d’un scanner abdominopelvien,
l’acquisition d’une unique série injectée au temps portal versus deux acquisitions identiques
sans et après injection au temps portal permet de réduire la dose de 50 %. La réalisation de
plusieurs temps d’acquisition doit donc être justifiée par une amélioration de la performance
diagnostique.

Le but de notre étude était d’évaluer la performance diagnostique de la série

abdominopelvienne sans injection lors de l’acquisition d’un scanner corps entier chez un
patient suspect de polytraumatisme.

Il s’agissait d’une étude monocentrique rétrospective réalisée au sein du service

d’imagerie Guilloz du CHRU de Nancy et incluant 84 scanners corps entier de patients
suspects de polytraumatisme. Deux lecteurs ont relu de manière indépendante les acquisitions
abdominopelviennes sans injection, avec injection et avec les deux séries sans et avec
injection. Une deuxième lecture par consensus représentait le gold standard. Lors de la lecture
des scanners, les lecteurs devaient dire s’il existait ou non une lésion traumatique
abdominopelvienne sur les différentes séries analysées. L’analyse statistique a porté sur la
concordance intra- et inter-observateur et sur la sensibilité et la spécificité des différentes
acquisitions par rapport à la relecture consensuelle.

Les résultats ne montraient pas de différence significative en terme de sensibilité ou de

spécificité pour la recherche d’une lésion traumatique entre l’interprétation de la série injectée
versus l’interprétation conjointe des deux séries sans et après injection. La concordance inter-
observateur était bonne à excellente. L’acquisition thoraco-abdominopelvienne en contraste
spontané représentait 20 % de la dose efficace de l’ensemble de l’examen.

Article 3 : Optimisation de la longueur d’acquisition des scanners réalisés pour colique

néphrétique : proposition d’une nouvelle méthode pour le placement de la limite
supérieure de l’acquisition.

44
 La réduction de la couverture d’acquisition est aussi un moyen simple, rapide et efficace
pour réduire la dose d’irradiation au scanner. Dans le cas des coliques néphrétiques,
l’acquisition abdominopelvienne peut être centrée sur les voies urinaires, du pôle supérieur
des reins au bord inférieur de la vessie. Tandis que la limite inférieure de l’acquisition est
facilement repérée par le milieu de la symphyse pubienne, le placement de la limite supérieure
de l’acquisition est plus compliqué. A l’HIA Legouest, nous utilisons la silhouette des reins
sur le topogramme de face afin de placer la limite supérieure de l’acquisition. Cette méthode
semble toutefois peu fiable avec de nombreux cas où le pôle supérieur d’un rein est coupé lors
de l’acquisition. Récemment, Corwin MT et al. [46] ont proposé un nouveau repère osseux
correspondant au bord inférieur de la dixième vertèbre thoracique (T10). Dans leur étude,
cette méthode permettait d’inclure dans tous les cas l’ensemble des reins. Par contre, il semble
que la réduction de la dose ne soit pas aussi importante qu’elle pourrait l’être. De ce fait, nous
proposons d’introduire un nouveau repère pour le placement de la limite supérieure de
l’acquisition qui correspond au point d’intersection entre la coupole diaphragmatique gauche
et le bord antérieur des corps vertébraux sur le scout de profil.

Le but de notre étude était de comparer ces trois méthodes de placement de la limite
supérieure de l’acquisition d’un scanner abdominopelvien en évaluant la réduction de la
couverture d’acquisition et le nombre de reins coupés.

Il s’agissait d’une étude rétrospective monocentrique incluant 365 scanners

abdominopelviens. Trois manipulateurs de radiologie ont placés pour chaque scanner la limite
supérieure de l’acquisition avec chacune des trois méthodes : méthode 1 utilisant les contours
des reins, méthode 2 utilisant le bord inférieur de la vertèbre T10 et méthode 3 utilisant le
point d’intersection entre la coupole diaphragmatique gauche et le bord antérieur des corps
vertébraux. La réduction de la couverture d’acquisition et le nombre de reins coupés ont été
comparés entre les trois méthodes.

Les résultats montraient une réduction moyenne de la longueur d’acquisition de 20,5 %,

15,1 % et 18,2 % avec respectivement les méthodes 1, 2 et 3. La proportion de reins coupés
pour les méthodes 1, 2 et 3 était de respectivement 6,7 %, 0,7 % et 1,4 %. Les concordances
inter-observateurs et intra-observateurs étaient excellentes pour toutes les méthodes mais les
coefficients de corrélation interclasse étaient toujours meilleurs pour la méthode 3.

45
46
 Chapitre 2

Article 1 : Evaluation des connaissances des prescripteurs de scanner en matière de

radioprotection des patients.

Gervaise A, Esperabe-Vignau F, Naulet P, Pernin M, Portron Y, Lapierre-Combes M. Evaluation des

connaissances des prescripteurs de scanner en matière de radioprotection des patients. J Radiol 2011; 92: 681-7.

47
48
Journal de radiologie (2011) 92, 681—687

ARTICLE ORIGINAL / Pratiques professionnelles

Évaluation des connaissances des prescripteurs de

scanner en matière de radioprotection des patients
Evaluation of the knowledge of physicians prescribing CT examinations on the
radiation protection of patients

A. Gervaise a,∗, F. Esperabe-Vignau b, M. Pernin a,

P. Naulet a, Y. Portron a, M. Lapierre-Combes a

a
Service d’imagerie médicale, hôpital d’instruction des armées Legouest, 27, avenue de
Plantières, BP 90001, 57077 Metz cedex 3, France
b
Service d’imagerie médicale, hôpital d’instruction des armées Robert-Picqué,
351, route de Toulouse, 33140 Villenave-d’Ornon, France

MOTS CLÉS Résumé

Dosimétrie ; Objectifs. — Évaluer les connaissances des praticiens prescripteurs de scanner en matière de
Prescripteur ; radioprotection des patients.
Radiation ionisante ; Matériels et méthodes. — Un questionnaire a été adressé à l’ensemble des praticiens pres-
Radioprotection ; cripteurs de scanner de notre hôpital. Ce questionnaire comportait plusieurs questions en
Scanner rapport avec les pratiques et les connaissances des praticiens en matière de radioprotection
des patients.
Résultats. — Quarante-quatre questionnaires ont été analysés. Tandis que 70 % des praticiens
déclaraient prendre en compte les risques liés aux rayons X lors de la prescription d’un scanner,
seulement 25 % de ceux-ci en informaient le patient. La connaissance des doses délivrées au
cours d’un scanner abdomino-pelvien était mal maîtrisée et les risques potentiels liés aux faibles
doses de rayons X étaient largement sous-estimés. Enfin, seulement un tiers des praticiens avait
bénéficié d’une formation à la radioprotection des patients.
Conclusion. — Même si la majorité des praticiens déclare prendre en compte les risques liés
aux faibles doses de rayons X, ces risques sont en fait peu ou mal connus. Une diffusion plus
large de la formation en radioprotection des patients, notamment au cours du cursus initial
des internes, pourrait être une des solutions pour améliorer les connaissances des praticiens
hospitaliers en matière de radioprotection des patients.
© 2011 Elsevier Masson SAS et Éditions françaises de radiologie. Tous droits réservés.

∗ Auteur correspondant.
Adresse e-mail : [Link]@[Link] (A. Gervaise).

0221-0363/$ — see front matter © 2011 Elsevier Masson SAS et Éditions françaises de radiologie. Tous droits réservés.
doi:10.1016/[Link].2011.03.023
682 A. Gervaise et al.

KEYWORDS Abstract
Dosimetry; Purpose. — To evaluate the knowledge of physicians prescribing CT examinations on the radia-
Ionizing radiation; tion protection of patients.
Radiation protection; Materials and methods. — A questionnaire was distributed to all clinicians on medical staff who
CT prescribe CT examinations. Several questions related to their prescription pattern and their
knowledge of radiation protection.
Results. — Forty-four questionnaires were analyzed. While 70% of physicians claimed that they
considered the risks from exposure to ionizing radiation when prescribing a CT examination,
only 25% informed their patients about those risks. Knowledge of the radiation dose delivered
during CT evaluation of the abdomen and pelvis was poorly understood and the risks related
to small doses of radiation were grossly underestimated. Finally, only a third of clinicians had
received training with regards to radiation protection.
Conclusion. — While most clinicians claim that they consider the risks from exposure to ionizing
radiation when prescribing a CT examination, the risks are either not well known or not known
at all. Increased formation of clinicians with regards to the radiation protection of patients,
maybe through a dedicated clinical rotation while in medical school, could be a solution to
improve the knowledge of hospital clinicians with regards to radiation protection.
© 2011 Elsevier Masson SAS and Éditions françaises de radiologie. All rights reserved.

En France, le recours aux examens d’imagerie médicale, européenne dans la directive Euratom 97/43 [12]. Cette
notamment scanographiques, est de plus en plus fréquent dernière précise que les examens d’imagerie irradiants
au cours de la prise en charge des patients. L’Institut de doivent faire l’objet d’une justification et d’une optimi-
radioprotection et de sûreté nucléaire (IRSN) rapporte ainsi sation constante afin de réduire les doses individuelles et
une augmentation de 26 % des actes de scanographie entre collectives dues aux expositions médicales.
2002 et 2007 [1]. Cela s’explique par une meilleure dispo- C’est dans ce contexte que la communauté radiolo-
nibilité des scanners, mais surtout par une amélioration gique (et plus largement, l’ensemble des professionnels de
constante de la qualité des images, dans un temps toujours santé utilisant directement des techniques d’imagerie irra-
plus court et pour une meilleure performance diagnostique. diantes) a été fortement sensibilisée ces dernières années.
Pourtant, le scanner est une technique d’imagerie irra- Sur le plan scientifique par exemple, ce n’est pas moins de
diante. Parallèlement à l’augmentation du nombre de 22 articles parus dans la revue Radiology en 2009 concernant
scanners réalisés chaque année, il existe également une le thème de la radioprotection ou de la réduction des doses
majoration de l’irradiation individuelle et collective déli- délivrées. C’est également une obligation de formation à
vrée au cours des actes de scanographie [2]. Ainsi, tandis la radioprotection des patients à renouveler tous les dix ans
qu’en 2007 les scanners ne représentaient que 10 % de pour les professionnels de santé utilisant les rayons X à visée
l’ensemble des actes de radiologie réalisés en France, diagnostique (arrêté du 18 mai 2004) [13]. Les radiologues
ils étaient responsables de 58 % de l’irradiation due aux sont aussi soumis à des niveaux de référence diagnostiques
rayons X d’origine médicale [1]. Or, même si le lien entre (c’est-à-dire des doses « seuil » par examen qu’il est souhai-
l’exposition à de faibles doses de rayons X et l’augmentation table de ne pas dépasser) définis par la législation (arrêté du
du risque de cancer radio-induit est fortement contro- 12 février 2004) [14] avec nécessité d’envoyer annuellement
versé, il a été établi par plusieurs grandes institutions des relevés de dosimétrie à l’IRSN. Enfin, dans l’optique de
(par exemple, le Comité scientifique des Nations-Unies et réduire au maximum les doses délivrées, notamment en sca-
l’Académie des sciences des États-Unis) et par de nombreux nographie, les radiologues doivent optimiser en permanence
rapports ou publications internationales [3,4]. En prenant les protocoles des examens irradiants.
en compte le modèle de régression linéaire sans seuil (RLSS) Tandis que de nombreux efforts ont été réalisés du côté
des effets des rayons X à faibles doses, certaines de ces de l’optimisation des examens irradiants, la justification de
publications n’hésitent d’ailleurs pas à associer à la réalisa- ces examens ne doit pas être négligée. Cette responsabilité
tion d’un seul scanner abdomino-pelvien un risque de cancer est partagée entre les praticiens prescripteurs et les radio-
radio-induit de l’ordre de 1/1000 [5,6]. Ce risque impor- logues. Elle impose le respect des indications, limitant la
tant reste toutefois un risque maximal théorique. D’autres réalisation des examens entraînant inutilement une exposi-
auteurs estiment pour leur part soit qu’un tel risque n’existe tion. Dans cette optique, le service d’imagerie médicale a
pas, soit qu’il est largement surestimé [7—11]. L’absence mis en ligne sur le réseau de notre hôpital un guide por-
de preuves scientifiques formelles allant dans un sens ou tant sur les indications des examens d’imagerie dans les
dans l’autre impose de suivre le principe de précaution urgences de l’adulte. Mais qu’en est-il du côté des pres-
et aboutit donc à la nécessité de réduire au maximum les cripteurs ? Plusieurs études de la littérature internationale
doses délivrées aux patients. Ce principe de précaution, « As insistent sur le manque d’implication des prescripteurs en
Low As Reasonable Achievable » : aussi bas que raisonnable- matière de radioprotection des patients ainsi que sur leur
ment possible (ALARA), a d’ailleurs été repris par l’Union ignorance en ce qui concerne les niveaux de doses délivrées
Évaluation des connaissances des prescripteurs de scanner 683

et les risques de cancer radio-induit actuellement admis Concernant la moyenne des doses délivrées au cours d’un
[15]. Aucune étude de ce type n’a jamais été réalisée en scanner abdomino-pelvien, le recueil dosimétrique a porté
France à notre connaissance. sur l’ensemble des scanners abdomino-pelviens réalisés au
L’objectif principal de notre étude était donc d’évaluer cours du mois d’avril 2010 (avec ou sans injection de produit
les connaissances en matière de radioprotection des de contraste, mono- ou multiphasique, scanner standard
patients des praticiens prescripteurs de scanner au sein de ou basse dose). Les doses délivrées étaient directement
notre hôpital. fournies par le rapport d’examen. Elles correspondaient
au produit dose longueur (PDL) exprimé en milliGray cen-
timètre (mGy × cm). La dose efficace (E), exprimée en
Matériels et méthodes millisievert (mSv) était ensuite calculée en utilisant le coef-
ficient de conversion tissulaire (k) de l’abdomen à 0,015 [16]
Participants selon la formule E = PDL × k [17].
Pour les radiographies thoraciques de face, l’estimation
Un questionnaire (Annexe 1) a été envoyé par courrier de la moyenne des doses délivrées a été réalisée à partir du
à l’ensemble des praticiens prescripteurs de scanner de recueil dosimétrique réalisé en 2009 dans le cadre de l’envoi
notre centre hospitalier la première semaine du mois d’avril annuel à l’IRSN des niveaux de référence diagnostiques. Ce
2010. Quatre-vingt-trois questionnaires ont ainsi été adres- recueil s’appuie sur les doses équivalentes, correspondant
sés à 60 seniors (médecins ou dentistes thésés) et 23 internes au produit dose surface (PDS) exprimé en Gy × cm2 , déli-
répartis dans les différents services de l’hôpital. vrées pour 20 radiographies thoraciques de face. La dose
efficace en mSv a été calculée à partir de la dose équivalente
en multipliant cette dernière par le coefficient de conver-
Questionnaire sion tissulaire kPDS (avec kPDS = 0,33 pour une radiographie
thoracique de face) selon la formule E = PDS × kPDS [18].
Ce questionnaire a été élaboré à partir des données de la
littérature, en concertation avec les différents médecins du
service d’imagerie médicale et de la personne compétente
en radioprotection. Résultats
Il comportait une première partie analysant les don-
nées démographiques du praticien (interne ou senior, années Estimation des doses délivrées
d’expériences depuis la thèse pour les seniors, service
d’appartenance). La moyenne des doses délivrées au cours d’un scanner
Le questionnaire était ensuite composé de huit ques- abdomino-pelvien était de 10,9 mSv et celle d’une radio-
tions qui abordaient plusieurs thèmes. Le premier thème graphie thoracique de face était de 0,07 mSv. Ces résultats
concernait les habitudes de prescription des praticiens : s’ils étaient en accord avec les niveaux de référence diagnos-
prescrivaient des scanners (question 1), s’ils avaient déjà tiques établis par l’IRSN [14] et avec les données de la
pris en compte le rapport bénéfice/risque lié aux rayons X littérature [5]. Le rapport entre la moyenne des doses d’un
lors de la prescription d’un scanner (question 2) et s’ils en scanner abdomino-pelvien et d’une radiographie thoracique
avaient déjà informé le patient (question 3). Le deuxième de face était donc de 155.
évaluait les connaissances des praticiens concernant les
doses délivrées au cours d’un scanner abdomino-pelvien
(questions 4 et 5) et des risques de cancer radio-induit du Résultats du questionnaire
fait de la réalisation de ce scanner (question 6). Enfin, la
Quarante-quatre questionnaires ont été récupérés et ana-
question 7 demandait aux praticiens s’ils avaient déjà suivi
lysés, soit un taux de réponse global de 53 % (65 % pour les
une formation à la radioprotection des patients tandis que
internes et 48 % pour les seniors). Ce sont donc 15 internes
la question 8 cherchait à savoir si les praticiens savaient
et 29 seniors qui ont répondu à notre questionnaire (soit
qu’il existe un guide portant sur les indications des examens
respectivement 34 et 66 % des effectifs de la population de
d’imagerie en urgence de l’adulte disponible sur le réseau
notre étude).
de l’hôpital.
Tous les praticiens ayant renvoyé le questionnaire étaient
prescripteurs de scanner.
Récupération et analyse des résultats Soixante-dix pour cent d’entre eux avaient répondu qu’ils
avaient déjà pris en compte le rapport bénéfice/risque lié
Le questionnaire devait être renvoyé au service d’imagerie aux rayons X lors de la prescription d’un scanner, avec un
médicale avant la fin du mois d’avril 2010. Les résultats ont pourcentage équivalent entre internes et seniors. Seule-
été analysés de manière anonyme. ment 25 % des praticiens en avaient déjà informé le patient.
C’est ainsi 31 % des seniors qui avaient déjà transmis une
Estimation des doses délivrées telle information au patient, contre seulement 13 % des
internes.
Afin d’établir la réalité des doses rapportées dans notre Concernant l’évaluation relative de la dose délivrée au
questionnaire, nous avons effectué un relevé dosimétrique cours d’un scanner abdomino-pelvien par rapport à une
au sein de notre service d’imagerie médicale concernant les radiographie thoracique de face, 13 % des praticiens avaient
doses délivrées au cours d’un scanner abdomino-pelvien et correctement évalué ce rapport entre 100 à 250 fois (Fig. 1).
d’une radiographie thoracique de face. Treize pour cent l’avaient surévalué au-delà de 250 tandis
684 A. Gervaise et al.

Par ailleurs, seulement 34 % des praticiens avaient déjà

bénéficié d’une formation à la radioprotection des patients.
L’analyse plus détaillée montrait qu’aucun interne n’avait
suivi de formation de ce type tandis que ce taux était de
52 % chez les seniors.
Enfin, 59 % des praticiens avaient connaissance de
l’existence d’un guide portant sur les indications des exa-
mens d’imagerie en urgence mis à leur disposition sur le
réseau de l’hôpital. Ce taux était plus élevé chez les internes
(67 %, contre 55 % chez les seniors).

Discussion
Notre étude confirme la mauvaise connaissance de la part
des praticiens des doses délivrées lors de la réalisation d’un
scanner abdomino-pelvien et la large sous-estimation du
Figure 1. Réponses à la question no 4 concernant le rapport entre
la dose délivrée au cours d’un scanner abdomino-pelvien standard
risque de cancer radio-induit qui en découle, compte tenu
(CT) et une radiographie thoracique de face (RT). des données actuelles de la littérature. Il est ainsi logique
qu’une part non négligeable d’entre eux ne prenne jamais
en compte le rapport bénéfice/risque lors de la prescrip-
que 74 % des praticiens avaient sous-estimé ce rapport, sans
tion d’un scanner et qu’une grande majorité des praticiens
distinction notable entre internes et seniors.
n’informe pas le patient de cette balance bénéfice/risque.
Lors de l’évaluation absolue de la dose délivrée au cours
Ce constat n’est pas isolé à notre établissement et est
d’un scanner abdomino-pelvien standard, avec pour repère
au contraire tout à fait concordant avec les données de la
l’irradiation naturelle en France estimée en moyenne à
littérature. De nombreuses études rapportent des résultats
2,5 mSv par an, 25 % des praticiens avaient correctement
équivalents [15].
évalué cette dose dans une fourchette allant de 5 à 20 mSv
Pour Lee et al., qui ont publié une étude comparable
(pour une dose moyenne de 10,9 mSv). Vingt-sept pour cent
en 2004 [19], seulement 9 % des urgentistes pensaient qu’il
des prescripteurs l’avaient surévaluée tandis que 48 % ne se
existait une augmentation du risque de cancer radio-induit
prononçaient pas ou avaient sous-estimé cette dose (Fig. 2).
du fait de la réalisation d’un seul scanner abdomino-pelvien
Tandis que l’estimation des doses délivrées était mal maî-
tandis que seulement 22 % de ceux-ci avaient déjà informé
trisée, les risques de cancer radio-induit étaient également
le patient d’un tel risque. Comme dans notre étude, le rap-
largement sous-estimés puisqu’une grande majorité des pra-
port de dose entre un scanner abdomino-pelvien et une
ticiens (61 % d’entre eux) avait répondu qu’il n’y avait aucun
radiographie thoracique de face était largement sous-estimé
risque de cancer radio-induit du fait de la réalisation d’un
(seulement 12 % des praticiens avaient correctement évalué
seul scanner abdomino-pelvien.
ou surévalué ce rapport).
Dans une autre étude publiée en 2004, Jacob et al.
[20] retrouvent également, dans une population de méde-
cins hospitaliers, une large sous-estimation du rapport de
dose entre un scanner abdomino-pelvien et une radiographie
thoracique de face (seulement 30 % des praticiens avaient
correctement estimé ce rapport) et une très large sous-
estimation du risque de cancer radio-induit du fait de la
réalisation d’un seul scanner abdomino-pelvien (12,5 % des
praticiens avaient évalué ce risque de manière correcte).
Plus surprenant, cette étude a montré que 10 % des pres-
cripteurs pensait que l’imagerie par résonance magnétique
(IRM) était aussi un examen d’imagerie irradiant ! Cette
étude mettait également en évidence un taux de réponses
correctes supérieur au sein du groupe de médecins ayant
déjà suivi une formation à la radioprotection des patients
comparativement au groupe n’ayant jamais suivi une telle
formation.
Enfin, dans une étude publiée en 2007 évaluant les
connaissances en matière de risque de cancer radio-induit
lié aux faibles doses de rayons X chez des chirurgiens pédia-
triques, Rice et al. [21] retrouvent que 31 % des praticiens
avaient correctement évalué le risque de cancer radio-induit
Figure 2. Réponses à la question no 5 concernant l’évaluation de du fait de la réalisation d’un seul scanner abdomino-pelvien
la dose délivrée au cours d’un scanner abdomino-pelvien (NSP : Ne tandis que 32 % estimaient que ce risque n’existait pas. Ces
sais pas). résultats plus favorables étaient, selon l’auteur, dus à deux
Évaluation des connaissances des prescripteurs de scanner 685

facteurs : tout d’abord, le fait qu’il s’agisse d’une étude (alors que les informations directement accessibles par les
réalisée chez des chirurgiens pédiatriques, population de patients n’abordent souvent que la notion de risque et non
médecins traditionnellement plus sensibilisée aux principes pas celle du bénéfice). Les prescripteurs doivent également
de la radioprotection, et deuxièmement, que cette étude connaître les modalités de calcul d’un tel risque et le débat
était plus récente que les deux précédentes et donc que actuel concernant sa probable surestimation.
ces meilleurs résultats pouvaient être expliqués par une Bien que ce ne soit pas le but de notre étude, plusieurs
meilleure diffusion des connaissances en matière de radio- précisions concernant les risques liés aux faibles doses de
protection envers les prescripteurs. rayons X méritent ainsi d’être mentionnées.
La formation des praticiens en matière de radioprotec- Ce risque est effectivement largement débattu au sein
tion des patients semble donc jouer un rôle important. Cette des communautés médicales et scientifiques. Malgré la
importance a été soulignée à la fois par les études de Rice controverse et les divergences de vues entre différentes
et Jacob mais également par la directive Euratom 97/43 qui institutions scientifiques reconnues au plan mondial, le
notait déjà en 1997 que « l’introduction d’un cours sur la modèle de RLSS est actuellement le modèle le plus large-
radioprotection dans le programme d’études de base des ment accepté, y compris pour de faibles doses d’irradiation.
facultés de médecine et d’art dentaire doit être favorisée » Le Comité scientifique des Nations-Unies pour l’étude des
[12]. Plus de dix ans après la diffusion de cette directive, effets des rayonnements ionisants (UNSCEAR) a déclaré
notre étude montre pourtant qu’aucune formation de radio- dans son rapport le plus récent [26] : « Tant que les [. . .]
protection des patients n’est enseignée au cours du cursus incertitudes sur les effets des faibles doses ne sont pas
initial des médecins. résolues, le Comité estime que la théorie selon laquelle
Deux principales raisons expliquent l’importance l’augmentation du risque de tumeur est proportionnelle
d’une telle formation auprès des praticiens prescripteurs à la dose d’irradiation est conforme aux connaissances
d’examens irradiants. Tout d’abord, en accord avec le prin- actuelles, et qu’elle reste donc l’approche la plus valable
cipe de précaution ALARA et dans l’optique de diminuer au d’un point de vue scientifique ». De l’autre côté, le rap-
maximum les doses délivrées aux patients, les prescripteurs port commun des experts de l’Académie française de
ont un rôle important quant à la justification des examens médecine et de l’Académie française des sciences conclut
d’imagerie irradiants. Cette responsabilité implique que pour sa part que le modèle de RLSS est incompatible
le prescripteur s’interroge sur le bénéfice par rapport au avec les dernières données scientifiques portant sur les
risque d’exposition, l’objectif étant d’obtenir l’information mécanismes de réparation moléculaire de l’ADN [7]. Les
diagnostique recherchée au moyen de la dose d’exposition auteurs de ce rapport estiment donc que même si le prin-
la plus faible. Toute irradiation, si faible soit-elle, doit être cipe de précaution est valable, il ne doit pas conduire
également justifiée par l’absence d’examen alternatif non à une surprotection superflue face à une surestimation
irradiant (notamment échographie ou IRM) [22]. des risques des faibles doses de rayons X. Ce risque doit
Deuxièmement, la formation des praticiens à la radiopro- plutôt être considéré comme un risque maximal théo-
tection vise à leur permettre de mieux informer les patients rique.
par rapport à la balance bénéfice/risque liée à la prescrip- Enfin, ce risque éventuel de cancer radio-induit doit être
tion d’un examen d’imagerie irradiant. Comme dans notre comparé au risque de cancer non radio-induit développé au
étude et dans les autres articles publiés [19,21], cette infor- sein de la population générale. Celui-ci est très élevé, de
mation reste très peu communiquée. L’apport systématique l’ordre de 42 % [3]. Cela signifie que si la réalisation d’un
au patient d’une telle information nous semble toutefois scanner engendre une augmentation du risque de cancer
excessif et discutable. La mention d’un risque de cancer radio-induit de l’ordre de 1/1000, le risque global de cancer
radio-induit ne doit pas dissuader le patient de recourir à passera donc de 42 à 42,1 %.
un examen qui s’avère souvent nécessaire et dont l’absence L’ensemble de ces données incite donc au principe de
peut être plus délétère qu’un risque hypothétique de cancer précaution sans pour autant surévaluer les risques de cancer
radio-induit. Toutefois, dans certaines circonstances, cette radio-induit associés aux faibles doses de rayons X.
information semble indispensable. C’est, par exemple, le Notre étude comporte plusieurs limites qui méritent éga-
cas particulier des femmes enceintes ou de jeunes patients lement d’être mentionnées. Tout d’abord, le faible effectif
présentant une pathologie chronique nécessitant la réalisa- de notre population d’étude ne nous a pas permis de
tion répétée de scanner (maladie de Crohn ou mucoviscidose séparer en différentes catégories les prescripteurs (par
par exemple). exemple, services médicaux contre chirurgicaux, années
Enfin, les médecins prescripteurs doivent s’attendre d’expérience. . .). De même, les différences entre internes
à répondre aux interrogations émanant directement des et seniors n’étaient pas statistiquement significatives, sauf
patients. En effet, de nombreuses informations sont doréna- en ce qui concerne le taux de formation à la radioprotec-
vant directement disponibles par le patient lui-même, que tion. La séparation entre praticien formé et non formé à la
ce soit dans la presse grand public ou sur internet [23,24]. radioprotection aurait également été un élément important
Il existe par exemple une application iPhone [25] et un pour souligner l’efficacité d’une telle formation.
site internet [24] permettant de calculer pour chaque type Par ailleurs, la volonté de faire un questionnaire pouvant
d’examen d’imagerie irradiant un risque de cancer radio- être rempli facilement et rapidement dans le but d’avoir un
induit. C’est dans ce contexte que le prescripteur doit être taux de réponse le plus élevé possible, ne nous a pas per-
capable de justifier la réalisation d’un examen irradiant mis d’approfondir certaines questions. Par exemple, il aurait
comportant un risque éventuel de cancer radio-induit. Le été intéressant de savoir à quelle occasion les praticiens
praticien doit notamment insister sur le bénéfice attendu de avaient bénéficié d’une formation à la radioprotection. Il
l’examen et sur le rapport bénéfice/risque qui en découle aurait aussi été intéressant de demander aux prescripteurs,
686 A. Gervaise et al.

comme dans l’étude de Jacob, s’ils pensaient que l’IRM était fait de la dose délivrée au cours d’un seul scanner abdomino-
un examen d’imagerie irradiant. pelvien standard :
Enfin, une diffusion plus large du questionnaire aurait été Oui/Non
intéressante afin de pouvoir comparer les connaissances en Question 7 :
matière de radioprotection au sein de notre population de Avez-vous déjà suivi une formation à la radioprotection
praticiens hospitaliers vis-à-vis d’une population de méde- des patients ?
cins généralistes. Oui/Non
Question 8 :
Savez-vous qu’il existe un guide portant sur les indi-
cations des examens d’imagerie en urgence de l’adulte
Conclusion
disponible sur le réseau de l’hôpital ?
Notre étude confirme que, même si la majorité des prati- Oui/Non
ciens déclare prendre en compte les risques liés aux faibles
doses de rayons X lors de la prescription d’un scanner,
ces risques sont en fait peu ou mal connus. Une diffusion Références
plus large de la formation en radioprotection des patients,
notamment au cours du cursus initial des internes, pour- [1] Etard C, Sinno-Tellier S, Aubert B. Exposition de la popula-
rait être une des solutions pour améliorer les connaissances tion française aux rayonnements ionisants liée aux actes de
des praticiens hospitaliers en matière de radioprotection des diagnostic médical en 2007. Rapport conjoint IRSN/InVS 2010.
patients. [2] National Council on Radiation Protection and Measurements.
Ionizing radiation exposure of the population of the United
States. NCRP report n◦ 160, 2009.
[3] Board of Radiation Effects Research Division on Earth and Life
Déclaration d’intérêts Sciences National Research Council of the National Academies.
Health Risks From Exposure to Low Levels of Ionizing Radiation:
Les auteurs déclarent ne pas avoir de conflits d’intérêts en BEIR VII Phase 2. Washington, DC: National Academies Press;
relation avec cet article. 2006.
[4] Brenner DJ, Doll R, Goodhead DT, Hall EJ, Land CE, Little JB,
et al. Cancer risks attributable to low doses of ionizing radia-
tion: assessing what we really know. Proc Natl Acad Sci U S A
Annexe 1. Questionnaire 2003;100:13761—6.
[5] Smith-Bindman R, Lipson J, Marcus R, Kim KP, Mahesh M, Gould
Question 1 : R, et al. Radiation dose associated with common computed
Prescrivez-vous des scanners ? tomography examinations and the associated lifetime attribu-
Oui/Non table risk of cancer. Arch Intern Med 2009;169:2078—86.
Question 2 : [6] Berrington de González A, Mahesh M, Kim KP, Bhargavan M,
Lewis R, Mettler F, et al. Projected cancer risks from computed
Lors de la prescription d’un scanner, avez-vous déjà pris
tomography scans performed in the United States in 2007. Arch
en compte le rapport bénéfice/risque lié aux rayons X ?
Intern Med 2009;169:2071—7.
Oui/Non [7] Tubiana M, Aurengo A, Averbeck A. La relation dose-effet
Question 3 : et l’estimation des effets cancérogènes des faibles doses de
Lors de la prescription d’un scanner, avez-vous déjà rayonnements ionisants. Paris: Académie nationale de méde-
informé le patient des risques liés aux rayons X et du rapport cine, institut de France — Académie des sciences; 2005.
bénéfice/risque qui en découle ? [8] Tubiana M, Feinendegen LE, Yang C, Kaminski JM. The linear
Oui/Non no-threshold relationship is inconsistent with radiation biologic
Question 4 : and experimental data. Radiology 2009;251:13—22.
Selon vous, comparativement à la dose délivrée pour une [9] Land CE. Low-dose extrapolation of radiation health risks:
some implications of uncertainly for radiation protection at
radiographie thoracique (RT) de face, la dose moyenne déli-
low doses. Health Phys 2009;97:407—15.
vrée au cours d’un scanner abdomino-pelvien standard (CT)
[10] Cordoliani YS. Parlons dose ou « ne dites pas à ma mère que je
équivaut à : suis radiologue : elle lit « Le Point ». J Radiol 2007;88:29—30.
CT < RT [11] Cordoliani YS. La dose efficace individuelle moyenne n’existe
10 RT > CT > RT pas. . . et pourtant elle augmente ! J Radiol 2010;91:449—50.
100 RT > CT > 10 RT [12] Directive 97/43/Euratom du 30 juin 1997, relative à la
250 RT > CT > 100 RT protection sanitaire des personnes contre les dangers des
CT > 250 RT rayonnements ionisants lors d’expositions à des fins médicales.
Question 5 : [13] Arrêté du 18 mai 2004 relatif aux programmes de formation
Sachant que l’irradiation naturelle en France est portant sur la radioprotection des patients exposés aux rayon-
nements ionisants. Journal officiel de la République française,
d’environ 2,5 mSv par an, à combien estimez-vous la dose
version consolidée du 26 septembre 2006.
moyenne délivrée au cours d’un scanner abdomino-pelvien ?
[14] Arrêté du 12 février 2004 relatif aux niveaux de référence diag-
.................................................................... nostiques en radiologie et médecine nucléaire. Journal officiel
.................................................................... de la République française, 16 mars 2004.
Question 6 : [15] Krille L, Hammer GP, Merzenich H, Zeeb H. Systematic review
D’après vous et selon les dernières conférences de on physician’s knowledge about radiation doses and radiation
consensus, existe-t-il un risque de cancer radio-induit du risks of computed tomography. Eur J Radiol 2010;76:36—41.
Évaluation des connaissances des prescripteurs de scanner 687

[16] Deak PD, Smal Y, Kalender WA. Multisection CT protocols: sex- [21] Rice HE, Frush DP, Harker MJ, Farmer D, Waldhausen JH. Peer
and age-specific conversion factor used to determine effective assessment of pediatric surgeons for potential risks of radia-
dose from Dose-Lengh-Product. Radiology 2010;257:158—66. tion exposure from computed tomography scans. J Pediatr Surg
[17] International Commission on Radiological Protection. 2007;42:1157—64.
2007 recommendations of the International Commission on [22] Haute Autorité de santé (HAS). Manuel de certification des
Radiological Protection (ICRP Publication 103). Ann ICRP établissements de santé. HAS; 2009.
2007;37:1—332. [23] Labbé C, Recasens O. Santé publique : les dangers de la radio-
[18] LeHeron J. Estimation of effective dose to the patient during logie. Le Point 2006.
medical X-ray examination from measurements of the dose [24] [Link]
area product. Phys Med Biol 1992;11:2117—26. [25] Radiation Passport. Tidal Pool Softwate. Version 2.0, 2010,
[19] Lee CI, Haims AH, Monico EP, Brink JA, Forman HP. Diagnos- [Link]
tic CT scans: assessment of patient, physician, and radiologist [26] UNSCEAR 2000 REPORT Vol. II: sources and effects of ionizing
awareness of radiation dose and possible risks. Radiology radiation. Annex G : biological effects at low radiation doses.
2004;231:393—8. Page 160, paragraph 541.
[20] Jacob K, Vivian G, Steel JR. X-Ray dose training: are we expo-
sed to enough? Clin Radiol 2004;59:928—34.
56
 Chapitre 2

Article 2 : Evaluation de l’intérêt de l’acquisition abdominopelvienne sans injection lors de

la réalisation d’un scanner corps entier chez un patient suspect de polytraumatisme.

Naulet P, Wassel J, Gervaise A, Blum A. Evaluation of the value of abdominopelvic acquisition without contrast
injection when performing a whole body CT scan in a patient who may have multiple trauma. Diagn Interv
Imaging 2013; 94: 410-7.

57
58
Diagnostic and Interventional Imaging (2013) 94, 410—417

ORIGINAL ARTICLE / Technical

Evaluation of the value of abdominopelvic

acquisition without contrast injection when
performing a whole body CT scan in a patient who
may have multiple trauma
P. Naulet a,∗, J. Wassel b, A. Gervaise a, A. Blum b

a
Medical Imaging Department, Hôpital d’Instruction des Armées Legouest, 27, avenue de
Plantières, BP 90001, 57077 Metz cedex 3, France
b
Guilloz Imaging Department, CHU Nancy, avenue de Lattre-de-Tassigny, 54000 Nancy, France

KEYWORDS Abstract
Multiple trauma; Purpose: To evaluate the diagnostic value of non-contrast-enhanced abdominopelvic acquisition
Abdominal trauma; when performing a whole body CT scan in a patient who may have multiple trauma.
Peritoneal effusion; Patients and methods: In a single centre, retrospective study over 1 year, we included 84
Multidetector CT patients suspected of having multiple trauma who indeed presented an abdominal or pelvic
scan; lesion during the initial CT scan. Two readers independently reread the acquisitions without
Iterative injection, then those with injection, then all the acquisitions, and scored the presence or
reconstruction absence of abdominopelvic lesions. Statistical analysis focused on intra- and inter-observer
agreement, and on the sensitivity and specificity of the different acquisitions in relation to
consensus rereading.
Results: This study did not reveal any significant difference, particularly concerning improve-
ment in sensitivity, between interpretation of the acquisitions with contrast injection and
interpretation of all the acquisitions with or without injection. Inter-observer agreement was
substantial to almost perfect. Non-contrast-enhanced thoraco-abdominopelvic acquisition rep-
resented 20% to 25% of the effective dose for the entire examination.
Conclusion: Abdominopelvic acquisition without contrast injection in addition to acquisition
with contrast injection in a patient suspected of having multiple trauma does not improve
detection of traumatic lesions of the liver, spleen, kidneys or adrenal glands, nor of intra- or
retroperitoneal effusion, but increases the dose and should be abandoned.
© 2013 Éditions françaises de radiologie. Published by Elsevier Masson SAS. All rights reserved.

Abbreviation: DLP, dose-length product.

∗ Corresponding author.
E-mail address: [Link]@[Link] (P. Naulet).

2211-5684/$ — see front matter © 2013 Éditions françaises de radiologie. Published by Elsevier Masson SAS. All rights reserved.
[Link]
Evaluation of CT acquisition without contrast injection in multiple trauma 411

A whole body CT scan has become the essential element looking for spontaneous hyperdensity resulting from the
in initial examination of a patient with suspected multiple presence of blood [13—15]. In particular, it is considered
trauma and for checking stable or stabilised haemodynamics of assistance in detecting small, particularly mesenteric
[1,2]. haematomas, haemoperitoneum and hepatic, splenic or
Performed early on, this examination provides an exhaus- renal haematomas. These lesions are hyperdense before
tive report of lesions and reduces mortality in multiple injection but are considered more difficult to detect after
trauma patients [3]. CT scanning protocols vary according to injection because of poorer contrast with the organs
the material available, the team’s habits and consideration enhanced (Figs. 1 and 2) [14]. Some teams undertake oral
of the X-ray dose delivered to the patient (although this is opacification [6,16], while others suggest only making acqui-
secondary where the patient’s state is critical). In partic- sitions centred on the region where trauma is suspected [12]
ular, non-contrast-enhanced abdominopelvic acquisition is but this attitude is controversial [17].
debated. The recommendations of the Société Française de Radi-
The literature suggests a number of protocols: most ologie (French Radiology Society) [18] for performing a CT
teams do not perform non-contrast-enhanced thoracic or scan in a patient with multiple trauma are: non-contrast-
abdominal acquisition [1,4—11], while others do undertake enhanced acquisition of the brain and neck, possibly
thoraco-abdominopelvic [12] or abdominal [13—15] acqui- non-contrast-enhanced thoraco-abdominopelvic acquisition
sition without contrast injection. This type of acquisition followed by thoraco-abdominopelvic acquisition (possibly
in the abdominal region is thought to be important for extending to the neck and legs) in the arterial phase (20

Figure 1. Mesenteric haemorrhagic contusion (white arrow) hyperdense with no contrast agent, showing a lesser degree of contrast with
the organs enhanced after injection: a: non-contrast-enhanced acquisition; b: acquisition after contrast injection in the portal phase.

Figure 2. Fracture of the left kidney with perirenal hyperdense haematoma with no injection of contrast agent (white arrows), appearing
hypodense after contrast injection but which can be characterised due to its density of 52 HU even with contrast-enhanced acquisition: a:
non-contrast-enhanced acquisition; b: acquisition after contrast injection in the portal phase.
412 P. Naulet et al.

to 30 s after starting the injection) then abdominopelvic The operators produced reformation of the various vol-
acquisition in the portal phase (70 to 90 s after the start umes in the three spatial planes as well as reformation of
of the injection) and finally, possibly, where there are renal the spine and aorta, then volume rendering reformation of
or perirenal anomalies or if there is any doubt about damage the thoracic cage, the face and bone lesions.
to the bladder, abdominopelvic acquisition in the late phase The resident and senior doctor analysed the results
(5 minutes). together on the PACS consoles (IMPAX V5, AGFA HealthCare),
In contrast, in its recommendations of June 2011 [19], or this was done first by the resident and then validated by
the British Royal College of Radiologists considers that non- the senior doctor.
contrast-enhanced thoraco-abdominopelvic acquisition is of
no interest in a traumatic context.
In our establishment, non-contrast-enhanced thoraco- Reading the CT scans
abdominopelvic acquisition is systematically performed in
patients suspected of having multiple trauma. In separate sessions at an interval of several days, reader
The aim of our study is to evaluate the diagnostic use- 1 (4th year resident) and reader 2 (senior doctor, registrar
fulness of this acquisition when performing a whole body CT in the department) independently reread only the non-
scan in a potentially multi-trauma patient. contrast injection abdominopelvic acquisitions then only the
abdominopelvic acquisitions with injection, and in a third
reading, all the abdominopelvic acquisitions with and with-
Patients and methods out contrast agent injection. The two readers then produced
a consensus rereading based on the results of all these
Population studied readings, the examination report recorded in the PACS, the
control scans and, for the 12 patients who had abdominal or
For this single centre, retrospective study, we searched our pelvic surgery, the operation reports.
establishment’s PACS for patients who had had an emergency For each series, they listed the presence or absence
whole body CT scan in the period between 01/01/2010 and of traumatic lesions of the liver, spleen or kidneys,
31/12/2010. of adrenal haematomas, of haemorrhagic intestinal-
This identified 282 patients who had had such a CT mesenteric lesions, of liquid peritoneal and retroperitoneal
scan for suspected multiple trauma. We reread all the effusion.
reports on these patients and included the 88 patients who The definitions used were those described in the paper
had at least one abdominopelvic traumatic lesion. Four by C. Ridereau-Zins et al. [14].
patients were excluded because their examination protocol All lesions were considered as being present whatever
was incomplete: one for whom the non-contrast-enhanced their size, severity and clinical significance. In particular,
thoraco-abdominopelvic acquisition was missing, three for the presence of a traumatic lesion of the liver, spleen
whom there had been no injection of an iodinated con- and kidneys was noted whenever there was a haematoma,
trast agent (one presenting a compressive acute subdural contusion, laceration or fracture. The intestinal-mesenteric
haematoma had required immediate neurosurgical treat- lesions noted were oedematous damage, haematomas and
ment, one had a history of allergy to iodinated contrast mesenteric haemorrhage, as well as digestive ischaemia sec-
agents and an 84-year-old patient had severe renal impair- ondary to these lesions and haematomas of the walls of
ment). the digestive tube. Peritoneal and retroperitoneal effusions
The only imaging examinations conducted prior to the were listed whatever their spontaneous density and abun-
whole body CT scan were frontal X-rays of the thorax in dance.
the resuscitation room, together with an ultrasound exam-
ination, in haemodynamically unstable patients, to detect
peritoneal, pericardial and pleural effusion. Statistical analysis
Technique for performing the CT scan The study population was subjected to a descriptive analy-
sis. The qualitative variables are shown as percentages and
All the examinations were performed in 64 × 0.5 mm heli- the quantitative variables are means with standard devia-
cal mode, 73 of them using a 320-row detector Aquilion tion.
One scanner (Toshiba Medical Systems, Tokyo, Japan) and 11 Cohen’s kappa coefficients, their standard error and the
with a 64-row detector Aquilion 64 scanner (Toshiba Medical adjusted kappa coefficients (PABAK) were calculated for
Systems, Tokyo, Japan). each lesion, comparing the results of non-contrast-enhanced
The protocol included producing non-contrast-enhanced acquisition, acquisition with injection and all acquisitions
cervico-encephalic and thoraco-abdominopelvic acquisi- with and without injection, in order to estimate intra- and
tions. After injection of 140 ml of contrast agent (iomeprol inter-observer agreement. Contingency tables were gener-
at 400 mg of iodine/ml, Bracco Altana Pharma, Constance, ated from the analyses performed.
Germany), acquisition was undertaken in the arterial phase To situate the kappa and PABAK coefficients obtained
extending from the base of the skull to the toes, fol- from our sample, we used the classification proposed by Lan-
lowed by abdominopelvic acquisition in the portal phase. dis and Koch: no agreement for negative values, slight from
Finally, if necessary, the resident or senior doctor present 0.0 to 0.20, fair from 0.21 to 0.40, moderate from 0.41 to
at the console decided whether to undertake a late 0.60, substantial from 0.61 to 0.80, almost perfect from 0.81
abdominopelvic acquisition. to 1 [20—23].
Evaluation of CT acquisition without contrast injection in multiple trauma 413

The sensitivity and specificity of each acquisition and the contrast injection. (The DLP of lower limb acquisition in
reports recorded in the PACS were calculated relative to the the arterial phase could not be separated from that of
consensus rereading. thoraco-abdominopelvic acquisition performed in the same
Data were entered using the Excel 2010 program from helix).
Microsoft Corporation (Redmond, Washington, USA) and the The effective dose (E) expressed in millisievert (mSv) was
statistical analysis used SAS® 9.2. (SAS Int. Inc., Cary, NC, then estimated, with the formula E = DLP × k, using a tissue
USA). conversion coefficient (k) of 18 mSv/mGy cm [24].
The statistical analysis was performed and the results
interpreted and presented with the help of an epidemi-
ologist from our establishment’s Clinical Epidemiology Results
Department.
The population studied consisted of 65 men (77%) and 19
Dosimetry women (23%) with a mean age of 38.8 years (standard devi-
ation of 17.8).
To study the additional dose of radiation delivered The prevalence of the various lesions is summarised in
to the patient during non-contrast-enhanced thoraco- Table 1.
abdominopelvic acquisition, we analysed the dosimetric Using the adjusted kappa (PABAK), intra-observer agree-
reports of the 73 examinations performed with the Aquil- ment between reading the acquisitions with injection and
ion One scanner. With the dosimetric reports of the 11 reading all the acquisitions varied depending on the lesions.
examinations performed with the Aquilion 64 scanner, the For reader 1, it was substantial to almost perfect (PABAK
dose-length product (DLP) for non-contrast-enhanced acqui- varying from 0.67 to 0.91). The agreement for reader 2 was
sitions could not be separated from the DLP for the also substantial to almost perfect (PABAK varying from 0.79
acquisitions with injection. to 0.91) (Table 2).
We calculated the mean and standard deviation of Inter-observer agreement was substantial to almost per-
the DLPs, expressed in [Link], for the non-contrast- fect for all the lesions for reading the acquisitions with
enhanced thoraco-abdominopelvic acquisitions and for the and without injection and for reading all the examinations,
thoraco-abdominopelvic and lower limb acquisitions with with the exception of moderate agreement for peritoneal

Table 1 Prevalence of lesions depending on the reader and acquisitions.

Organs Reader 1 Reader 2 Consensus

NI INJ NI + INJ NI INJ NI + INJ

Adrenal haematoma 7 (8.3) 12 (14.3) 13 (15.5) 9 (10.7) 13 (15.5) 11 (13.1) 13 (15.5)
Liver lesion 2 (2.4) 22 (26.2) 22 (26.2) 4 (4.8) 26 (31) 25 (29.8) 25 (29.8)
Spleen lesion 5 (6) 24 (28.6) 23 (27.4) 8 (9.5) 23 (27.4) 24 (28.6) 25 (29.8)
Kidney lesion 5 (6) 12 (14.3) 10 (11.9) 6 (7.1) 12 (14.3) 11 (13.1) 13 (15.5)
Intestinal-mesenteric lesion 5 (6) 7 (8.3) 9 (10.7) 15 (17.9) 13 (15.5) 14 (16.7) 11 (13.1)
Peritoneal effusion 34 (40.5) 55 (65.5) 55 (65.5) 46 (54.8) 53 (63.1) 56 (66.7) 57 (67.9)
Retroperitoneal effusion 28 (33.3) 31 (36.9) 41 (48.8) 34 (40.5) 39 (46.4) 32 (38.1) 38 (45.2)
Number of cases (%).
NI: acquisition without contrast injection; INJ: acquisition with contrast injection; NI + INJ: all acquisitions; Consensus: consensus
rereading.

Table 2 Intra-observer agreement (acquisitions with contrast injection vs. all acquisitions).
Organs Reader 1 Reader 2

K (SE) 95% CI PABAK K (SE) 95% CI PABAK

Adrenal haematoma 0.67 (0.11) 0.45—0.89 0.83 0.81 (0.09) 0.62—0.99 0.91
Liver lesion 0.88 (0.06) 0.76—0.99 0.91 0.86 (0.06) 0.74—0.98 0.88
Spleen lesion 0.85 (0.06) 0.73—0.98 0.88 0.79 (0.08) 0.65—0.94 0.83
Kidney lesion 0.79 (0.10) 0.59—0.99 0.91 0.75 (0.11) 0.54—0.96 0.88
Intestinal-mesenteric lesion 0.44 (0.16) 0.13—0.77 0.81 0.69 (0.11) 0.48—0.91 0.83
Peritoneal effusion 0.89 (0.05) 0.79—0.99 0.91 0.79 (0.07) 0.62—0.91 0.79
Retroperitoneal effusion 0.67 (0.08) 0.51—0.82 0.67 0.78 (0.07) 0.65—0.92 0.79
K: Kappa; SE: Standard error; CI: confidence interval.
414 P. Naulet et al.

effusion when interpreting non-contrast-enhanced acquisi- Discussion

tions (PABAK = 0.48) (Table 3).
There was no improvement in sensitivity between reading In our study, intra-observer agreement between the acquisi-
acquisitions with injection and reading all the acquisitions tions with injection and all the acquisitions was substantial
with and without injection, except in the case of retroperi- to almost perfect. In addition, the sensitivity of acquisitions
toneal effusion by reader 1. with injection was not improved by interpreting acquisitions
Sensitivity in detecting liver and spleen lesions was sig- without injection at the same time.
nificantly improved through interpreting the acquisitions There is therefore no advantage to performing non-
with contrast injection compared with just the non-injection contrast-enhanced acquisition in addition to acquisition
acquisitions. A similar tendency was found for kidney with contrast injection in patients with one or more
lesions, adrenal haematomas and peritoneal effusions. abdominopelvic traumatic lesions.
Sensitivity in detecting mesenteric lesions and retroperi- This corresponds with the recommendations [19] and
toneal effusion did not seem to be improved by reading the practice of very many teams who do not under-
either the acquisitions with or without injection (Table 4). take non-contrast-enhanced abdominopelvic or thoraco-
Specificity was excellent for the two readers and all the abdominopelvic acquisitions in patients with suspected
acquisitions. multiple trauma [1,4—11].
Considering intestinal-mesenteric haemorrhagic lesions In 1988, Kelly J. et al. [13] showed the usefulness of
more particularly, only two patients had an emergency oper- non-contrast-enhanced abdominal acquisition in addition
ation, one for a haematoma of the greater omentum with to contrast-enhanced acquisition for abdominal trauma. In
active haemorrhage, the other for a wound to a branch their study, the sensitivity and specificity of all the exam-
of the mesenteric artery caused by a penetrating trauma inations were improved from 74 to 92% and from 84 to
(metal bar transfixing the pelvis, abdomen and thorax, in 91% respectively, by acquiring 10 slices of 10 mm spaced at
position when the CT scan was performed). Three other 20 mm without contrast injection on the upper part of the
patients underwent surgery for suspected perforation of abdomen.
the digestive tube: the first indeed presented perforation In 1992, a study by Miyakawa K. et al. [25] confirmed
of the small intestine, the second perforation of the blad- these results in 126 patients. Non-contrast-enhanced acqui-
der (false-positive), while for the third no perforation was sition helped in particular to diagnose the 12 traumatic
found, although he had a pneumoperitoneum. The other intestinal lesions requiring emergency surgical manage-
intestinal-mesenteric lesions were mesenteric haemorrha- ment, whereas with acquisition with contrast injection only
gic contusions (5 cases) or small haematomas (2 cases) which 10 were diagnosed.
did not require surgery and diagnosis of which could not be Since these old studies, however, CT technology and
confirmed. image quality have considerably improved. Volume acqui-
In our series, the DLP for non-contrast-enhanced thoraco- sitions are nowadays interpreted on PACS consoles with
abdominopelvic acquisition was a mean of 1499 [Link] multiplanar reformations.
(standard deviation: 482), i.e. 24% of that of all the thoraco- Our study’s inter-observer agreement is clearly much
abdominopelvic and lower limb acquisitions, which was higher than in the study by Agostini et al. [11] on the
6256 [Link] (standard deviation: 1953). The mean of the usefulness of dual reading of whole body CT scans in the
effective dose for acquisition without contrast injection was management of multiple trauma patients, which found an
approximately 27 mSv. inter-observer kappa of 0.41 (95% confidence interval of 0.35

Table 3 Inter-observer agreement.

Organs Acquisitions without injection Acquisitions with injection Acquisitions with and
without injection
K (SE) 95% CI PABAK K (SE) 95% CI PABAK K (SE) 95% CI PABAK
Adrenal 0.59 (0.15) 0.29—0.88 0.86 0.77 (0.10) 0.57—0.96 0.88 0.81 (0.09) 0.62—0.99 0.91
haematoma
Liver lesion 0.66 (0.23) 0.21—1 0.95 0.83 (0.07) 0.69—0.96 0.86 0.79 (0.07) 0.65—0.94 0.83
Spleen lesion 0.75 (0.14) 0.48—1 0.93 0.73 (0.08) 0.57—0.89 0.79 0.73 (0.08) 0.57—0.89 0.79
Kidney lesion 0.90 (0.09) 0.71—1 0.98 0.71 (0.11) 0.49—0.93 0.86 0.62 (0.13) 0.36—0.88 0.83
Intestinal- 0.23 (0.13) −0.03—0.49 0.67 0.33 (0.15) 0.04—0.61 0.71 0.45 (0.14) 0.18—0.72 0.74
mesenteric
lesion
Peritoneal 0.49 (0.09) 0.31—0.66 0.48 0.85 (0.06) 0.73—0.96 0.86 0.71 (0.08) 0.55—0.87 0.74
effusion
Retroperitoneal 0.75 (0.08) 0.59—0.89 0.76 0.61 (0.09) 0.44—0.78 0.62 0.64 (0.08) 0.48—0.80 0.64
effusion
K: Kappa; SE: Standard error; CI: confidence interval.
Evaluation of CT acquisition without contrast injection in multiple trauma 415

Table 4 Sensitivity and specificity of lesions depending on the reader and acquisitions.
Reader 1 Reader 2

NI INJ NI + INJ NI INJ NI + INJ

Adrenal haematoma
Sensitivity 0.54 (0.27) 0.85 (0.20) 0.77 (0.23) 0.54 (0.27) 0.85 (0.20) 0.85 (0.20)
[0.27—0.81] [0.65—1.00] [0.54—1.00] [0.27—0.81] [0.65—1.00] [0.65—1.00]
Specificity 1.00 (0.00) 0.99 (0.03) 0.96 (0.05) 0.97 (0.04) 0.97 (0.04) 1.00 (0.00)
[1.00—1.00] [0.96—1.00] [0.91—1.00] [0.93—1.00] [0.93—1.00] [1.00—1.00]
Liver lesion
Sensitivity 0.08 (0.11) 0.84 (0.14) 0.88 (0.13) 0.16 (0.14) 0.96 (0.08) 0.92 (0.11)
[0.00—0.19] [0.70—0.98] [0.75—1.00] [0.02—0.30] [0.88—1.00] [0.81—1.00]
Specificity 1.00 (0.00) 0.98 (0.03) 1.00 (0.00) 1.00 (0.00) 0.97 (0.05) 0.97 (0.05)
[1.00—1.00] [0.95—1.00] [1.00—1.00] [1.00—1.00] [0.92—1.00] [0.92—1.00]
Spleen lesion
Sensitivity 0.20 (0.16) 0.88 (0.13) 0.88 (0.13) 0.32 (0.18) 0.88 (0.13) 0.88 (0.13)
[0.04—0.36] [0.75—1.00] [0.75—1.00] [0.14—0.50] [0.75—1.00] [0.75—1.00]
Specificity 1.00 (0.00) 0.97 (0.05) 0.98 (0.03) 1.00 (0.00) 0.98 (0.03) 0.97 (0.05)
[1.00—1.00] [0.92—1.00] [0.95—1.00] [1.00—1.00] [0.95—1.00] [0.92—1.00]
Kidney lesion
Sensitivity 0.31 (0.25) 0.92 (0.14) 0.77 (0.23) 0.38 (0.26) 0.77 (0.23) 0.77 (0.23)
[0.06—0.56] [0.78—1.00] [0.54—1.00] [0.12—0.65] [0.54—1.00] [0.54—1.00]
Specificity 0.99 (0.03) 1.00 (0.00) 1.00 (0.00) 0.99 (0.03) 0.97 (0.04) 0.99 (0.03)
[0.96—1.00] [1.00—1.00] [1.00—1.00] [0.96—1.00] [0.93—1.00] [0.96—1.00]
Intestinal-mesenteric lesion
Sensitivity 0.36 (0.28) 0.55 (0.29) 0.64 (0.28) 0.73 (0.26) 0.64 (0.28) 0.64 (0.28)
[0.08—0.65] [0.25—0.84] [0.35—0.92] [0.46—0.99] [0.35—0.92] [0.35—0.92]
Specificity 0.99 (0.03) 0.99 (0.03) 0.97 (0.04) 0.90 (0.07) 0.92 (0.06) 0.90 (0.07)
[0.96—1.00] [0.96—1.00] [0.94—1.00] [0.84—0.97] [0.85—0.98] [0.84—0.97]
Peritoneal effusion
Sensitivity 0.54 (0.13) 0.96 (0.05) 0.96 (0.05) 0.79 (0.11) 0.93 (0.07) 0.91 (0.07)
[0.41—0.67] [0.92—1.00] [0.92—1.00] [0.68—0.90] [0.86—1.00] [0.84—0.99]
Specificity 0.89 (0.12) 1.00 (0.00) 1.00 (0.00) 0.96 (0.07) 1.00 (0.00) 0.85 (0.13)
[0.77—1.00] [1.00—1.00] [1.00—1.00] [0.89—1.00] [1.00—1.00] [0.72—0.99]
Retroperitoneal effusion
Sensitivity 0.71 (0.14) 0.76 (0.14) 0.97 (0.05) 0.82 (0.12) 0.89 (0.10) 0.79 (0.13)
[0.57—0.85] [0.63—0.90] [0.92—1.00] [0.69—0.94] [0.80—0.99] [0.66—0.92]
Specificity 0.98 (0.04) 0.96 (0.06) 0.91 (0.08) 0.93 (0.07) 0.89 (0.09) 0.96 (0.06)
[0.94—1.00] [0.90—1.00] [0.83—1.00] [0.86—1.00] [0.80—0.98] [0.90—1.00]
NI: acquisition without contrast injection; INJ: acquisition with contrast injection; NI + INJ: all acquisitions.
Value (standard error) [95% confidence interval].

to 0.46) for all the lesions of a whole body CT scan. That can or little consequence (e.g. simple contusions or small sub-
be explained by our studying the agreement lesion by lesion, capsular haematomas of solid organs, adrenal haematomas
and only being concerned with the abdomen. We also used and small peritoneal or retroperitoneal effusions) and only
the PABAK, which sometimes differs from kappa. required monitoring. A false-positive or false-negative for
On the other hand, in the study by Yu J. et al. [16] on these lesions had no impact on management of the patient.
isolated small peritoneal effusions, the kappa coefficient for Conversely, serious haemorrhagic lesions are life-
this single sign between two observers was 0.76, which gives threatening and require immediate surgery and intensive
a result close to the inter-observer kappa that we found for care. When they concern the liver, spleen or kidneys, their
free peritoneal effusions (kappa = 0.85 for the acquisitions diagnosis poses no problem from just contrast-enhanced
with injection and 0.74 for all acquisitions). acquisitions.
An important limitation of our study was taking into On the other hand, mesenteric and intestinal trauma
account all the lesions visible on the CT scan whatever their is rare, particularly serious and more difficult to diag-
size and clinical impact. Many of these lesions were of no nose. Its early diagnosis depends almost exclusively on the
416 P. Naulet et al.

abdominal CT scan because clinical signs and symptoms are Since optimisation of the protocol and the use of
non-specific. A false-negative can result in delayed diagno- iterative reconstruction, the DLP and therefore the effec-
sis responsible for increased morbidity and mortality due to tive dose, which is proportional to the DLP, have been
haemorrhage, sepsis and peritonitis [6,26—29]. reduced by about 70% for non-contrast-enhanced thoraco-
A limitation of our study was that it included only very abdominopelvic acquisition and by about 64% for the entire
few lesions of this type that had required surgical treat- protocol.
ment (one mesenteric haematoma with active bleeding; one Since these modifications, the additional dose of
haemorrhage due to a penetrating trauma — but detection radiation (8 mSv) due to non-contrast-enhanced thoraco-
of this lesion posed no problem since the object causing the abdominopelvic acquisition performed at low dose and
injury was still in situ; two bowel perforations, one of which read in semi-thin slices has been very significantly reduced
was not found by surgery; and a false-positive for bladder compared with the former protocol (27 mSv), but it still rep-
perforation). The seven other lesions only required monitor- resents about 20% of the dose of the whole protocol and
ing. eliminating it would mean a further dose reduction.
It could be useful to conduct an additional study focusing The limitations of our study are due to its retrospec-
solely on traumatic intestinal-mesenteric lesions that have tive and single centre character and, as we saw earlier,
required surgical management. This additional study would to analysis of all lesions without consideration for either
help overcome the limitations of our work for these rare but their severity or their therapeutic impact. However, even
serious lesions that are difficult to diagnose. frequent diagnostic errors on lesions without clinical con-
There could be two disadvantages to undertak- sequences are less serious than a single error which is
ing non-contrast-enhanced thoraco-abdominopelvic or life-threatening.
abdominopelvic acquisition during a whole body CT scan in Moreover, even when taking into account operation
a patient suspected of having multiple trauma: the time reports and clinical evolution, consensus rereading is a
taken and the irradiation. source for discussion, particularly concerning the presence
The time for acquisition is less than 10 s, even count- or absence of minimal lesions which have no therapeutic
ing the time for programming it, moving the table, etc. impact, but may have a statistical impact by modifying the
The increase in time for the patient in the CT scan room sensitivity and specificity of the different readings. During
is 1 to 2 min. Positioning the patient and the topograms consensus rereading, a minor lesion was recorded if it had
are the same as for the contrast-enhanced acquisitions been found by one reader on one acquisition but missed by
and therefore do not lengthen the protocol. Reconstruct- the other reader or on other acquisitions. This explains a low
ions are done during the preparation for and while carrying number of false-positives and thus the excellent specificity
out the contrast-enhanced acquisitions and do not increase of all the acquisitions. To limit these biases, it would have
the length of the examination. The time taken to perform ideally been best to list the severity of the different lesions.
this acquisition is therefore negligible compared with the Finally, our study was limited to abdominopelvic lesions
patient’s total stay in the CT scan room, which is about to the exclusion of studying thoracic lesions. This choice
30 min. was made because among the 282 patients suspected of
Our evaluation of the effective dose is not very precise having multiple trauma and who underwent a whole body
and has numerous biases: the DLP not taking into account CT scan, only four presented rupture or dissection of the
either the length of exploration or the patient’s morpho- aortic isthmus necessitating treatment and two presented
type, the DLP of acquisition in the arterial phase including doubtful untreated lesions which remained stable on the
the legs, and use of a single tissue conversion coefficient control scans. Moreover, only one presented a haemoperi-
(k). cardium. The total number was considered too small to
Since performing our study, the protocol has been opti- be studied. In addition, the presence of a mediastinal
mised, while retaining all the acquisitions, with considerable haematoma, detection of which could possibly be improved
reduction in the exposure parameters and the use of iter- by non-contrast-enhanced acquisition, shows few specific
ative reconstruction algorithms. We therefore studied the differences from a large vessel lesion [1].
dosimetry of the first 20 patients for the month of January
2012 who were scanned using the Aquilion One scanner with
a whole body protocol for suspected multiple trauma. Conclusion
During these examinations, the mean DLP for non-
The disadvantages of performing non-contrast-enhanced
contrast-enhanced thoraco-abdominopelvic acquisition was
abdominopelvic acquisition in addition to acquisition with
449 [Link] (standard deviation: 136), i.e. 20% of the DLP
contrast injection in a patient suspected of having multiple
of all the thoraco-abdominopelvic and lower limb acquisi-
trauma are loss of time, which is minimal, but an increase
tions, which was 2282 [Link] (standard deviation: 799).
in dose of about 20 to 25%. It does not improve detection
On average, the effective dose for non-contrast-enhanced
of traumatic lesions of the liver, spleen, kidneys or adrenal
acquisition was about 8 mSv but represented a little more
glands, nor of intra- or retroperitoneal effusion, and should
than 20% of the total effective dose. Indeed, the total
be abandoned.
effective dose is slightly overestimated due to the use
of a global tissue conversion factor without separating
thoraco-abdominopelvic acquisition in the arterial phase Disclosure of interest
from acquisition of the lower limbs. However, the tissue
conversion factor for the lower limbs is much lower than The authors declare that they have no conflicts of interest
the global tissue conversion factor used. concerning this article.
Evaluation of CT acquisition without contrast injection in multiple trauma 417

Acknowledgements [13] Kelly J, Raptopoulos V, Davidoff A, Waite R, Norton P. The value

of non-contrast-enhanced CT in blunt abdominal trauma. AJR
We warmly thank Mrs Irawati Lemonnier, MPH, PhD, of the Am J Roentgenol 1989;152:41—8.
[14] Ridereau-Zins C, Lebigot J, Bouhours G, Casa C, Aubé C.
CIC-CE Inserm, and of the Epidemiology and Clinical Evalua-
Traumatismes abdominaux : les lésions élémentaires. J Radiol
tions Department of Nancy University Teaching Hospital for
2008;89:1812—32.
her assistance. [15] Taourel P. Imagerie des urgences. Paris: Masson; 2004,
p. 211—34.
[16] Yu J, Fulcher AS, Wang DB, Turner MA, et al. Frequency
References and importance of small amount of isolated pelvic free fluid
detected with multidetector CT in male patients with blunt
[1] Taourel P, Merigeaud S, Millet I, Devaux Hoquet M, Lopez trauma. Radiology 2010;256(3):799—805.
FM, Sebane M. Traumatisme thoraco-abdominal : stratégie en [17] Tillou A, Gupta M, Baraff LJ, Schriger DL, Hoffman JR,
imagerie. J Radiol 2008;89:1833—54. Hiatt JR, et al. Is the use of pan-computed tomography for
[2] Miller LA, Shanmuganathan K. Multidetector CT evalua- blunt trauma justified? A prospective evaluation. J Trauma
tion of abdominal trauma. Radiol Clin N Am 2005;43: 2009;67(4):779—87.
1079—95. [18] Guide pratique à l’usage des médecins radiologues pour
[3] Wagner SH, Lefering R, Qvick L-M, Körner M, Kay MV, Pfeifer l’évaluation de leur pratiques professionnelles. Paris: Société
K-J, et al. Effect of whole-body CT during trauma resuscita- Française de Radiologie; 2009. p. 610—2.
tion on survival: a retrospective, multicentre study. Lancet [19] The Royal College of Radiologists. Standards of practice and
2009;373:1455—61. guidance for trauma radiology in severely injured patients.
[4] Salim A, Sangthong B, Martin M, Brown C, Plurad D, Demetri- London: The Royal College of Radiologists; 2011.
ades D. Whole body imaging in blunt trauma patient without [20] Landis JR, Koch GG. The measurement of observer agreement
obvious sign of injury: results of a prospective study. Arch Surg for categorical data. Biometrics 1977;33:159—74.
2006;141:468—73. [21] Blum A, Feldmann L, Bresler F, Jouanny P, Briançon S, Régent D.
[5] Holmes JF, McGahan JP, Wisner D. Rate of intra-abdominal Intérêt du calcul du coefficient kappa dans l’évaluation d’une
injury after a normal abdominal computed tomographic scan méthode d’imagerie. J Radiol 1995;76(7):441—3.
in adults with blunt trauma. Am J Emerg Med 2012;30(4): [22] Sim J, Wright CC. The Kappa statistic in reliability studies:
574—9. use, interpretation, and sample size requirements. Phys Ther
[6] Brofman N, Atri M, Epid D, Hanson JM, Grinblat L, et al. Evalua- 2005;85:257—68.
tion of bowel and mesenteric blunt trauma with multidetector [23] Kundel HL, Polansky M. Measurement of observer agreement.
CT. Radiographics 2006;26:1119—31. Radiology 2003;228:303—8.
[7] Holly BP, Steenburg SD. Multidetector CT of blunt traumatic [24] Huda W, Ogden KM, Khorasani MR. Converting dose-
venous injuries in the chest, abdomen, and pelvis. Radiograph- length product to effective dose at CT. Radiology
ics 2011;31:1415—24. 2008;248(3):995—1003.
[8] Anderson SW, Varghese JC, Lucey BC, Burke PA, Hirsch EF, Soto [25] Miyakawa K, Kaji T, Ashida H, Kuwabara M, et al. Evaluation of
JA. Blunt splenic trauma: delayed-phase CT for differentiation non-contrast-enhanced CT in blunt abdominal trauma. Nippon
of active hemorrhage from contained scular injury in patients. Acta Radiol 1992;52(3):300—7.
Radiology 2007;243:88—95. [26] Fakhry SM, Brownstein M, Watts DD, Baker CC, Oller D. Rela-
[9] Nguyen D, Platon A, Shanmuganathan K, Mirvis SE, Becker tively short diagnostic delays (8 hours) produce morbidity and
CD, Poletti P-A. Evaluation of a single-pass continuous whole- mortality in blunt small bowel injury: an analysis of time to
body 16-MDCT protocol for patients with polytrauma. AJR operative intervention in 198 patients from a multicenter expe-
2009;192:3—10. rience. J Trauma 2000;48(3):408—14 [discussion 414—5].
[10] Briggs RH, Rowbotham E, Johnstone AL, Chalmers AG. Pro- [27] Le Bedis CA, Anderson SW, Soto JA. CT imaging of blunt trau-
visional reporting of polytrauma CT by on-call radiology matic bowel and mesenteric injuries. Radiol Clin North Am
registrars. Is it safe? Clin Radiol 2010;65:616—22. 2012;50(1):123—36.
[11] Agostini C, Durieux M, Milot L, Kamaoui I, Floccard B, Allaouch- [28] Wu CH, Wang LJ, Wong YC, Fang JF, et al. Contrast-
iche B, et al. Intérêt de la double lecture du scanner corps enhanced multiphasic computed tomography for identifying
entier dans la prise en charge des polytraumatisés. J Radiol life-threatening mesenteric hemorrhage and transmural bowel
2008;89:325—30. injuries. J Trauma 2011;71(3):543—8.
[12] Gupta M, Schriger DL, Hiatt JR, Cryer HG, Tillou A, et al. Selec- [29] Tan KK, Liu JZ, Go TS, Vijayan A, Chiu MT. Computed tomo-
tive use of computed tomography compared with routine whole graphy has an important role in hollow viscus and mesenteric
body imaging in patients with blunt trauma. Ann Emerg Med injuries after blunt abdominal trauma. Injury Int J Care Injured
2011;58:407—16. 2010;41:475—8.
 Chapitre 2

Article 3 : Optimisation de la longueur d’acquisition des scanners réalisés pour colique

néphrétique : proposition d’une nouvelle méthode pour le placement de la limite supérieure
de l’acquisition.

Gervaise A, Teixeira P, Hossu G, Blum A, Lapierre-Combes C. Optimizing z-axis coverage of abdominal CT

scans of the urinary tract: a proposed alternative proximal landmark for acquisition planning. En cours de
soumission à British Journal of Radiology.

67
68
 ”‹‰‹ƒŽ‡•‡ƒ” Š
Optimizing z-axis coverage of abdominal CT scans of the urinary tract:
a proposed alternative proximal landmark for acquisition planning.
Gervaise A, Teixeira P, Hossu G, Blum A, Lapierre-Combes M.

Submitted to British Journal of Radiology on February 2016.

Abstract
Objective: To evaluate an alternative method to Introduction
reduce the acquisition coverage of urinary tract Since its introduction in the 1990s, unenhanced
CT. Computed Tomography (CT) has become the
Materials and Methods: This retrospective gold standard for urinary tract imaging in
study included 365 abdominopelvic CT studies. patients with renal colic [1, 2]. It offers many
Three technologists simulated shortened advantages: availability, no injection of iodinated
acquisition coverages using three methods to contrast medium and excellent diagnostic
determine the upper limit of the acquisition: performance [3]. Its main limitation is related to
method 1 used the renal contours; method 2 used radiation, especially as urinary stone disease
the inferior margin of the 10th thoracic vertebra; mainly affects young patients with a tendency to
and method 3 used the point of intersection of relapse [4, 5]. In accordance with the cautionary
the left diaphragmatic dome and the anterior principle of ALARA (As Low As Reasonably
margin of the vertebral bodies. Reductions in Achievable) and given the potential risk of
acquisition coverage and number of cut kidneys radiation-induced cancers associated with low X-
were compared between the three methods. ray doses [6, 7], reducing the dose urinary tract
Results: The mean reduction of acquisition CT is paramount.
coverage for the three readers with methods 1, 2 Many studies have shown that it is possible to
and 3 were, 20.5%, 15.1% and 18.2%, use low-dose CT to investigate renal colic with
respectively. The proportions of cut kidneys with excellent diagnostic performance [3]. The
methods 1, 2 and 3 and averaged over the three researchers primarily evaluated the possibility of
readers were 6.7%, 0.7% and 1.4%, respectively. reducing the dose by increasing the pitch [8],
Inter and intra-reader agreement was excellent lowering tube current [9-11], using automatic
with all methods, but inter-class correlation tube current modulation [12], reducing tube
coefficients were higher with method 3. voltage [13, 14] and using iterative
Conclusion: Using the intersection of the left reconstruction algorithms [13-19].
diaphragmatic dome and the anterior margin of Another simple and effective way to reduce the
the vertebral bodies for proximal landmark for dose is to limit acquisition coverage from the top
urinary tract CT is more reproducible than of the kidneys to the lower edge of the bladder.
conventional methods and reduces by 18.2% the For the lower acquisition limit, the lower edge of
acquisition coverage without significantly the symphysis pubis bone is an accurate
increasing kidneys cuts. landmark [20, 21]. Identifying the upper limit of
the acquisition is more complicated. In our
Keywords: Computed Tomography; radiation institution, we reduce the acquisition coverage
dosage; anatomical landmark; scan coverage; by trying to locate the top of the kidneys on the
urinary tract. frontal scout image. However, the renal contours
is frequently hard to identify and to our
A. Gervaise – Corresponding author, M. Lapierre-Combes knowledge, the effectiveness of this method has
Service d’Imagerie Médicale, Hôpital d’Instruction des Armées
Legouest, 27 avenue de Plantières, 57070 Metz, CEDEX 3, France. not yet been evaluated.
E-mail: [Link]@[Link] In order to find a more reliable method of
reducing acquisition coverage without cutting the
P. Teixeira, A. Blum kidneys, two recent studies have evaluated the
Service d’Imagerie Guilloz, Hôpital Central, CHU-Nancy, 29 Av.
Mar De Lattre de Tassigny, 54035 Nancy, France. possibility of using a vertebral bony landmark to
define the upper acquisition limit. For de Leon et
G. Hossu al. [20] use of the superior margin of 11th
Université de Lorraine, IADI, UMR 947, Tour Drouet, rue du thoracic vertebral body (T11) reduced the
Morvan, 54511 Vandoeuvre-lès-Nancy, France.

 ͳ
 Gervaise et al.
acquisition coverage by 18%, compared to Body Mass Index (BMI) was calculated using
standard abdominopelvic acquisition, without the formula BMI = weight / height2 (kg / m2).
cutting the kidneys. For Corwin et al., [21]
placing the upper limit of the acquisition at the CT acquisition and reconstruction parameters:
inferior margin of the 10th thoracic vertebral All examinations were performed in supine
body (T10) led to a 17.7% coverage reduction position using a 64-slice multidetector CT
without cutting the kidneys. Although these two scanner (OPTIMA CT660, General Electric
studies showed that the use of T10 or T11 bony Healthcare, USA). Examinations began with a
landmark excluded no kidneys during the frontal and lateral scout radiograph with 120 kV
acquisition, we believe that the reduction of the and 10 mA during a deep in-breath. Acquisition
scan coverage is not so important it could be in of both frontal and lateral scout radiographs was
many patients. In addition, the relatively high necessary to enable automatic exposure control.
frequency of anatomical variants in the thoraco- The examinations included at least one
lumbar spine and the fact that it is not applicable abdominopelvic acquisition at the portal phase
in patients with scoliosis adds to the difficult in from the top of diaphragm through the ischial
applying these methods in clinical practice. tuberosities. In some cases additional series were
We propose an alternative method for placement performed (unenhanced, arterial or delayed
of the upper acquisition limit of urinary tract CT- phases). These series were not evaluated in this
scans. This method is based on the lateral scout study. Acquisition parameters included tube
image and uses the point of intersection between voltage from 100-140 kV according to weight,
the left diaphragmatic dome and the anterior automatic tube current modulation (GE Smart
margin of the vertebral bodies. We believe that mA) with a noise index of between 18 and 25
this may allow optimal coverage reduction while and ASIR (Adaptive Statistical Iterative
using a simple anatomical landmark, which Reconstruction). Images were reconstructed in
could represent an advantage over conventional thin slices of 1.25 mm every 1.25 mm with a soft
methods. The aim of our study was to compare tissue kernel. All examination images as well as
three coverage-reducing methods with an the scout images and the review report with
evaluation of scan coverage reduction percentage location data for the first and last slices were
and number of cut kidneys. archived in our PACS database (Picture
Archiving and Communication System).

Materials and Methods Z-axis scan coverage evaluation:

This study was approved by our local ethics For each CT-scan, the coverage of the standard
committee. Taking into account its retrospective abdominopelvic acquisition was calculated as the
nature, written consent of patients was not difference in cm between the location of the first
necessary. and last slices of the acquisition.
The location of the lower limit of the simulated
Study population: reduced acquisition was standardized for all
This retrospective single-center study included three methods and was identical to that used by
all patients over 18 years old referred to our de Leon et al. and Corwin et al. [20, 21]. It
imaging department between November 2014 corresponded to the section passing through the
and February 2015 for a standard inferior aspect of the bony symphysis and was
abdominopelvic CT-scan with injection of placed on each standard abdominopelvic CT-
iodinated contrast medium. Four patients were scan by the study investigator (A.G.), a
excluded because the frontal or lateral scout radiologist with 9 years of clinical experience,
image did not cover the entire abdomen, one using a post-processing workstation ADW 4.6
because of severe scoliosis and three because of (GE HealthCare, USA).
left pulmonary effusion that precluded The location of the upper limit of the simulated
visualization of the left diaphragmatic dome on reduced acquisition was different for each of the
the lateral scout image. Overall, 365 CT-scans three methods:
involving 343 patients were included in the Method 1 corresponded to the placement of the
study. upper limit of the acquisition from the frontal
For all patients, the following parameters were scout image by locating the top of the kidneys.
systematically collected at the time of the scan: Method 2 was that described by Corwin et al.
sex, age (years), weight (kg) and height (m). The [21], and corresponded to the placement of the
upper limit of the acquisition from the frontal

 ʹ
 Gervaise et al.
scout image at the level of the inferior margin of location on the z-axis on the scout images.
T10. The lowest rib-bearing vertebral body was Readers were able to zoom and change the
considered to represent the 12th thoracic vertebral window settings of scout images. A second
body. Method 3 was the alternative method we reading session was conducted by readers 2 and
propose. With this method, the upper limit of the 3, two months after the first reading on 30 CT-
acquisition was placed on the lateral scout image scans randomly selected and placed in a different
at the point of intersection between the anterior random order to the first session. This second
margin of the vertebral body and the left reading was used to calculate intra-observer
diaphragmatic dome (Fig 1). The left dome was agreement.
identified by the lowest position of the right Having determined the upper limits, the lengths
dome. In case of doubt, the frontal scout image of simulated reduced acquisition were calculated
was used to confirm the laterality of the dome. for each scan and each method as the difference
For each scan, the upper limits of the acquisition in cm between the location of the upper and
for all three methods were determined by three lower limits.
CT technologists with at least 4 years of The study investigator (A.G.) noted, for each
experience in scanning for renal colic in our standard abdominopelvic CT, the z-axis location
institution using method 1. With regard to the of the superiormost aspect of the highest kidney.
upper limit selection, technologists had access to By comparing the location data of the upper limit
frontal and lateral scout images, which had been of the simulated reduced acquisition and the top
anonymized and randomized. The upper limit of the uppermost kidney, it was possible to
was placed on ADW 4.6 workstation (GE determine how many kidneys were cut by each
Healthcare, USA) using the "localizer" function, method and for each reader and evaluate the
which allowed the determination of the slice length of the kidney portions that were cut out.

Fig. 1 - Frontal (a, b) and lateral (c) scout images showing how to place the upper limit of the reduced acquisition with
the three methods evaluated. For method 1 (a), the upper limit of the acquisition was placed by locating the top of the
kidneys using the renal contours as reference (arrowheads). Method 2 (b) used the inferior margin of T10 as reference.
Method 3 (c) used the lateral scout image and the point of intersection (arrow) between the anterior margin of the
vertebral bodies (solid white line) and the left diaphragmatic dome (dashed white line).

Statistical analysis: respectively, reduced acquisition coverage and

Data were analyzed using R software for proportion of cut kidney among the three
Windows Version 3.2.0 (R Foundation for methods. An interclass correlation coefficient
Statistical Computing, Vienna, Austria). The (ICC) was used to assess the inter-reader
values were given as the mean ± standard agreement and intra-reader agreement for
deviation with the extreme values in parentheses. technologists 2 and 3. A p value less than 0.05
For each of the three technologists, the Tukey was considered to represent a statistically
test and Fisher's exact test were used to compare, significant difference.

 ͵
 Gervaise et al.

Table 1. Simulated acquisition coverage reduction according to each reader and each of the three methods.

Simulated acquisition coverage reduction (cm) p value

M1 M2 M3 M2/M1 M3/M1 M3/M2
Reader 1 35.77 ± 2.94 37.25 ± 2.03 36.01 ± 3.10 <0.001 0.483 <0.001
Reader 2 34.10 ± 2.89 37.45 ± 2.08 35.91 ± 3.04 <0.001 <0.001 <0.001
Reader 3 34.96 ± 2.82 37.15 ± 2.12 35.87 ± 3.04 <0.001 <0.001 <0.001
M1 = method 1; M2 = method 2; M3 = method 3.
Results given as mean ± standard deviation

Results Table 2. The percentage of cut kidneys with

methods 1, 2 and 3, averaged over the three
Study population: readers, were respectively 6.7%, 0.7% and 1.4%.
Of the 343 patients enrolled, 178 were women Method 1 resulted in most kidney cuts, with a
and 165 men with an average age of 54.7 years significant variation between the three readers
(range 18-92 years), a mean weight of 77.2 kg (from 7 for reader 1 to 46 for reader 2). There
(35 to 170 kg), an average height of 167.9 cm was no significant difference between method 2
(143 to 199 cm) and a mean BMI of 27.2 kg / m² with respect to method 3, regardless of the reader
(14.9 to 60 kg / m²). (P> 0.451). For method 2, three same studies had
a cut kidney with reader 1 and 3 and two of these
Reduced simulated acquisitions: three studies had a cut kidney with reader 2.
The average coverage of standard
abdominopelvic acquisition was 43.93 ± 3.81 cm
(33.87 to 54.75 cm). The scan coverage of the
reduced simulated acquisitions for each of the
three methods and all three technologists are
shown in Table 1 and Fig 2. For all
technologists, method 1 permitted a greater
reduction in acquisition coverage over standard
abdominopelvic acquisition, and method 2
produced the smallest reduction in acquisition
coverage. For readers 2 and 3, the differences
between the three methods were statistically
significant (p <0.001). For reader 1, only the
difference between methods 1 and 3 was not
statistically significant (p = 0.483).
Compared to standard abdominopelvic
acquisition, the percentage of reduction in
acquisition coverage with methods 1, 2 and 3,
averaged over the three readers, were,
respectively 20.5%, 15.1% and 18.2 %.
The average distance between the lower limit of Fig. 2 - Chart showing the distribution of simulated
standard abdominopelvic acquisition and the reduced acquisition coverages according to the three
methods for each of the three readers. Note that with
inferior aspect of the symphysis pubis was 2.3 methods 2 and 3, the acquisition coverages are very
cm, which corresponded to a 5.2% reduction of consistent between the three readers in comparison
the length of acquisition. Reduction of the with the method 1. Also note the greater reduction in
acquisition coverage linked to the upper limit the acquisition coverage with method 3 compared to
alone was 15.2%, 9.9% and 13% for methods 1, method 2.
2 and 3, respectively.

Number of cut kidneys: The maximum length of cut out kidney portions
The numbers of CT-scans with a cut kidney for with methods 1, 2 and 3 was respectively 4.2 cm,
each method coverage optimization are shown in 0.6 cm and 1 cm for the three readers

 Ͷ
 Gervaise et al.
(respectively 2.1 cm, 0.5 cm and 1 cm for reader In order not to cut kidneys, the bony landmark
1; 4.2 cm, 0.5 and 1 cm for reader 2; 2.8 cm, 0.6 should be the middle of the vertebral body of
cm and 0.8 cm for reader 3). T10.

Table 2. Number of CT-scans with a portion of kidney excluded from the simulated reduced acquisition
based on each reader and each of the three methods.

Number of CT-scans with a portion of kidney excluded from the

p value
simulated reduced acquisition
M1 M2 M3 M2/M1 M3/M1 M3/M2
Reader 1
7 (1.9 %) 3 (0.8 %) 5 (1.4 %) 0.772 0.882 0.772
Reader 2
46 (12.6 %) 2 (0.5 %) 5 (1.4 %) <0.001 <0.001 0.451
Reader 3
20 (5.5 %) 3 (0.8 %) 5 (1.4 %) 0.001 0.005 0.725
M1 = method 1; M2 = method 2; M3 = method 3.
Results given as number and percentage in parentheses.

Intra- and inter-reader agreements: agreements of readers 2 and 3 were excellent,

The inter-reader agreement was excellent regardless of the method and the reader (Table
regardless of the method (Table 3). There was a 4). For reader 3, the intra-reader agreement was
progressive increased in ICC values from significantly better for methods 2 and 3 than for
method 1 to 3, method 3 having the overall best method 1.
inter-reader agreement. The intra-reader

Table 3: Inter-reader agreement between the three methods.

Method ICC 95 % CI

M1 0.926 0.814 < ICC < 0.962

M2 0.979 0.974 < ICC < 0.983

M3 0.992 0.991 < ICC < 0.994

CI = Confidence Interval; ICC = Interclass Correlation Coefficient;
M1 = method 1; M2 = method 2; M3 = method 3.

Table 4: Intra-reader agreement for readers 2 and 3 based on 30 CT-scans randomly selected.
Readers Method ICC 95 % CI
M1 0.937 0.073 < ICC < 0.985
Reader 2 M2 0.986 0.972 < ICC < 0.994
M3 0.982 0.962 < ICC < 0.991
M1 0.984 0.962 < ICC < 0.993
Reader 3 M2 0.999 0.998 < ICC < 1
M3 1 1 < ICC < 1
CI = Confidence Interval; ICC = Interclass Correlation Coefficient; M1 = method 1; M2 = method 2; M3 = method 3

 ͷ
 Gervaise et al.

Discussion Leon et al. [20] and Corwin et al. [21], our study
shows that the superior margin of T10 is not
Our study confirms that a 15-20% reduction in infallible and that in order not to cut kidneys, the
the coverage of urinary tract CT-scans can be bony landmark should be the middle of the
obtained by optimizing the acquisition protocol. vertebral body of T10. However, with such a
As it has been shown for other CT applications landmark, reduced acquisition coverage would
[22, 23], reducing the acquisition coverage is a be increased for roughly 1 cm. In one particular
simple and effective way of limiting the dose in patient with small 12th ribs hardly visible on the
patients and is recommended in clinical practice. scout image, readers 2 did not cut the kidney,
Because of the close relation between scan because the inferior margin of the 9th thoracic
length and effective dose, when the acquisition vertebral body was mistakenly used as landmark
length is reduced a similar reduction in effective (Fig 3). This shows that the vertebral bony
dose can be expected. In a recent study assessing landmark method can also have pitfalls.
the possibility to reduce the CT scan coverage
for acute appendicitis which was also based on ƒ
T10 anatomical landmark, Corwin et al. found
that a reduction of the scan coverage of 24 %
corresponded to a dose reduction of 23 % [22].
In addition, by optimizing coverage, it is also
possible to reduce the exposure of radiosensitive
organs, such as the gonads in men and the
breasts in women [24]. In practice, like others
dose reduction techniques such as iterative
reconstruction algorithms [13-19, 25],
optimization of the acquisition coverage could
have a significant impact on the dose delivered
to patients undergoing a urinary tract CT.
One of the major findings of our study is that the
use of the kidney’s contours as reference on the
frontal scout image is difficult in practice
(method 1). Even if it reduces the acquisition
coverage by up to 20.5% compared to a standard
abdominopelvic acquisition, it was the least „
reproducible and the most likely method to cut
the upper pole of the kidney. For reader 2, in
12% of cases the kidneys were cut with a
maximal distance of cut kidney of 4.2 cm, which
is not acceptable in clinical practice. This
limitation is related to the difficulty of
visualizing the kidney on the frontal scout image.
These findings underscore the importance of
identifying a more suitable anatomical landmark
for the determination of the upper limit of
urinary tract CT acquisitions [20, 21].
In our study, we used the bony landmark of
Corwin et al., which corresponded to the inferior Fig. 3 - Female aged 65 years with a BMI of 26.4 kg /
margin of T10 (method 2) [21] and not the m². Frontal scout image (a) with green lines
superior margin of T11 as de Leon et al. [20]. corresponding to the limits of the standard
Although the difference between these two abdominopelvic acquisition and red line corresponding
landmarks is minimal, we chose to study the to the upper limit of the reduced acquisition coverage
uppermost landmark in order to favor the method using method 2. The axial CT slice (b) passing
that excluded the least kidneys. While for through the lower margin of T10 shows that the top of
the left kidney is cut (arrow). Also note the presence
Corwin et al, 100% of kidneys were included in
of small ribs T12 (arrowhead) that were not seen by
the reduced acquisition, in our study, readers 1 reader 2, who mistakenly used the inferior margin of
and 3 excluded the upper pole of the left kidney T9 as landmark and didn’t cut the left kidney.
in three patients using method 2. In contrast to de

 ͸
 Gervaise et al.
The proposed method (3), based on the lateral Moreover, according to current clinical practice,
scout image and taking as a reference point the a patient whose scan was negative would
intersection of the anterior margin of the probably undergo an additional standard
vertebral bodies and the left diaphragmatic dome enhanced CT to search for possible differential
may be of interest to optimize the coverage of diagnoses. Fourth, detail and clarity of scout
urinary tract CT-scans and thereby reduce the images may vary across different CT vendors.
delivered dose. It was the most reproducible The results of our study should be confirmed on
method of the three, both between readers and CT-scanners from others manufacturers. Finally,
for an individual reader. It is not hampered by we used a CT protocol with automatic tube
anatomical variants and uses landmarks that are current modulation, which implies that the dose
quick and easy to find, which is supported by reduction is not strictly proportional to the
excellent inter- and intra-reader agreement. In reduction in acquisition coverage. Therefore, we
particular, the distinction between the left and did not directly estimate the percentage reduction
right diaphragmatic dome on the lateral scout in the dose.
image poses no problem because the left In conclusion, using the intersection of the left
diaphragmatic dome is often lower than the right. diaphragmatic dome and the anterior margin of
It is also possible to detect its position relative to the vertebral bodies on the lateral scout image to
the right diaphragmatic dome on the frontal scout set the upper limit of the acquisitions of CT-
image. This method allows a significant scans in patients with suspected renal colic
reduction of the acquisition coverage over appears to be a good compromise. This method
method 2 (18.2 against 15.1% p <0.001). allows for a greater reduction of acquisition
Although the proportion of kidneys cut using coverage in comparison with the inferior margin
method 3 was greater than with method 2 (1.4% T10 landmark and excludes fewer kidneys than
vs 0.7%), this difference was not statistically the method using kidney contours on frontal
significant. In addition, with method 3, in the scout image. It is also the most reproducible
CT-scans with cut kidneys, less than 1 cm of method evaluated that we now use in clinical
kidney was out of the field of view, which seems practice.
acceptable in patients evaluated for renal colic.
This method strikes as a good compromise
between method 1 that cuts too many kidneys Acknowledgements:
with a greater reduction in coverage, and method None.
2, which rarely cuts kidneys but gives a smaller
coverage and dose reduction.
Our study has several limitations. First, it is References
retrospective. A prospective study with a larger
patient population would be interesting to 1. Coursey CA, Casalino DD, Remer EM,
confirm our results, especially given the low Arellano RS, Bishoff JT, Dighe M, et al.
number of CT-scans with a cut kidney using our ACR Appropriateness Criteria® acute onset
new method. Second, we included all patients flank pain: suspicion of stone disease.
who had a standard abdominopelvic CT, without Ultrasound Q 2012;28:227-233.
selecting those referred for suspected renal colic. 2. Kambadakone AR, Eisner BH, Catalano
This allowed us to include more patients and OA, Sahani DV. New and evolving concepts
compare reductions in acquisition length to the in the imaging and management of
length of standard abdominopelvic acquisition in urolithiasis: urologists’ perspective.
the same patient. However, three patients had to Radiographics 2010;30:603-623.
be excluded from our study because they had left 3. Niemann T, Kollmann T, Bongartz G.
pulmonary effusion that masked the left Diagnostic performance of low-dose CT for
diaphragmatic dome on the scout image and detection of urolithiasis: a meta-analysis.
prevented us using the proposed method. In AJR Am J Roentgenol 2008;191; 396-401.
clinical practice, this has little impact since the 4. Broder J; Bowen J, Lohr J Babcock A, Yoon
association of left sided pleural effusion and J. Cumulative CT exposures in emergency
renal colic is relatively rare. Third, we did not try department patients evaluated for suspected
to assess the frequency of lesions (e.g. renal colic. J Emerg Med 2007; 33:161-168.
pulmonary nodule, adrenal mass) that might be 5. Katz SI, Saluja S, Brink JA, Forman HP.
missed on acquisitions with reduced coverage, as Radiation dose associated with unenhanced
it was not among the objectives of this study. CT for suspected renal colic: impact of

 ͹
 Gervaise et al.
repetitive studies. AJR Am J Roentgenol 15. Fontarensky M, Alfidja A, Perignon R,
2006;186: 1120-1124. Schoenig A, Perrier C, Mulliez A, et al.
6. Smith-Bindman R, Lipson J, Marcus R, Lim Reduced radiation dose with Model-based
KP, Mahesh M, gould R, et al. Radiation Iterative Reconstruction versus standard
dose associated with common computed dose with Adaptive Statistical Iterative
tomography examinations and the associated Reconstruction in abdominal CT for
lifetime attributable risk of cancer. Arch diagnosis of acute renal colic. Radiology
Intern Med 2009;169:2078-2086. 2015;276:156-66.
7. Berrington de González A, Mahesh M, Kim 16. McLaughlin PD, Murphy KP, Hayes SA,
KP, Bhargavan M, Lewis R, Mettler F, et al. Carey K, Sammon J, Crusch L, et al. Non-
Projected cancer risks from computed contrast CT at comparable dose to an
tomography scans performed in the United abdominal radiograph in patients with acute
States in 2007. Arch Intern Med renal colic; impact of iterative
2009;169:2071-2077. reconstruction on image quality and
8. Diel J, Perlmutter S, Venkataramanan N, diagnostic performance. Insights Imaging
Mueller R, Lane MJ, Katz DS. Unenhanced 2014;5:217-230.
helical CT using increased pitch for 17. Wang J, Kang T, Arepalli C, Barrett S,
suspected renal colic: an effective technique O’Connell T, Louis L, et al. Half-dose non-
for radiation dose reduction? J Comput contrast CT in the investigation of
Assist Tomogr 2000;24:795-801. urolithiasis: image quality improvement
9. Tack D, Sourtzis S, Delpierre I, de with third-generation integrated circuit CT
Maertelaer V, Gevenois PA. Low-dose detectors. Abdom Imaging 2015;40:1255-
unenhanced multidetector CT of patients 62.
with suspected renal colic. AJR Am J 18. Glazer DI, Maturen KE, Cohan RH,
Roentgenol 2003: 180: 305-311. Davenport MS, Ellis JH, Knoepp US, et al.
10. Kluner C, Hein PA, Gralla O, Hein E, Assessment of 1 mSv urinary tract stone CT
Hamm B, Romano V, et al. Does ultra-low- with Model-Based Iterative Reconstruction.
dose CT with a radiation dose equivalent to AJR Am J Roentgenol 2014;203:1230-1235.
that of KUB suffice to detect renal and 19. Winklehner A, Blume I, Winklhofer S,
ureteral calculi? J Comput Assist Tomogr Eberli D, Gnannt R, Frauenfelder T, et al.
2006; 30:44-50. Iterative reconstructions versus filtered
11. Poletti PA, Platon A, Rutschmann OT, back-projection for urinary stone detection
Schmidlin FR, Iselin CE, Becker CD. Low- in low-dose CT. Acad Radiol 2013;20:1429-
dose versus standard-dose CT protocol in 1435.
patients with clinically suspected renal colic. 20. de Leon AD, Xi Y, Champine J, Costa DN.
AJR Am J Roentgenol 2007; 188; 927-933. Achieving ideal computed tomography scan
12. Mulkens TH, Daineffe S, De Wijngaert R, length in patient with suspected urolithiasis.
Bellinck P, Leonard A, Smet G, et al. J Comput Assist Tomogr 2014;38:264-267.
Urinary stone disease: comparison of 21. Corwin MT, Bekele W, Lamba R. Bony
standard-dose and low-dose with 4D MDCT landmarks on computed tomography
tube current modulation. AJR Am J localizer radiographs to prescribe a reduced
Roentgenol 2007;188:553-562. scan range in patients undergoing
13. Kulkarni NM, Uppot RN, Eisner BH, Sahani multidetector computed tomography for
DV. Radiation dose reduction at suspected urolithiasis. J Comput Assist
multidetector CT with Adaptive Statistical Tomogr 2014;38:404-7.
Iterative Reconstruction for evaluation of 22. Corwin MT, Chang M, Fananapazir G,
urolithiasis: how low can we go? Radiology Seibert A, Lamba R. Accuracy and radiation
2012;265:158-166. dose reduction a limited abdominopelvic CT
14. Gervaise A, Naulet P, Beuret F, Henry C, in the diagnosis of acute appendicitis.
Pernin M, Portron Y, et al. Low-dose CT Abdom Imaging 2015;40:1177-82.
with automatic tube current modulation, 23. Campbell J, Kalra MK, Rizzo S, Maher
adaptive statistical iterative reconstruction MM, Shepard JA. Scanning beyond
and low tube voltage for the diagnosis of anatomic limits of the thorax in chest CT:
renal colic: impact of body mass index. AJR findings, radiation dose and automatic tube
Am J Roentgenol 2014;202:553-560. current modulation. AJR Am J Roentgenol
2005;185:1525-30.

 ͺ
 Gervaise et al.

24. van der Molen AJ, Geleijns J. Overranging 25. Gervaise A, Osemont B, Lecocq S, Noel A,
in multisection CT: quantification and Micard E, Felblinger J, et al. Image quality
relative contribution to dose - comparison of improvement using Adaptive Iterative Dose
four 16-section CT scanners. Radiology Reduction with wide-volume acquisition on
2007;242:208-16. 320-detector CT. Eur radiol 2012;22:295-
301.

 ͻ
78
Discussion et conclusion du chapitre 2 :

Les résultats de nos études mettent en avant l’importance des facteurs comportementaux
dans une démarche d’optimisation et de réduction de la dose d’irradiation au scanner.

Article 1 : Evaluation des connaissances des prescripteurs de scanner en matière de

radioprotection des patients.

Les résultats de notre étude confirment qu’il existe un défaut de connaissance des
praticiens hospitaliers prescripteurs de scanner vis-à-vis des niveaux de dose de rayons X
délivrés au cours d’un scanner abdominopelvien et surtout vis-à-vis des risques potentiels de
cancer radio-induit lié aux faibles doses de rayons X actuellement admis par la communauté
scientifique. Bien que le nombre de réponse soit relativement faible (seulement 44
questionnaires analysés dont 29 séniors et 15 internes), ces résultats sont concordants avec les
données de la littérature [44]. Cette étude montre toutefois que 70 % des praticiens prennent
en compte les risques liés aux rayons X lors de la prescription d’un scanner. Ce résultat est
plutôt positif car cela suggère que plus de la moitié des praticiens est sensibilisée à cette
problématique. Or, cette sensibilisation des prescripteurs est un élément important pour
pouvoir faire appliquer les principes de justification et de substitution des scanners. Etant
donné que seulement un tiers des prescripteurs avait bénéficié d’une formation à la
radioprotection des patients, la mise en place de formation pour les médecins au cours des
études de médecine et au cours de la formation médicale continue semble être une des
solutions pour mieux diffuser ce type de connaissance et pour améliorer la radioprotection des
patients. D’ailleurs, suite aux résultats de cette étude, nous avons mis en place une séance de
formation de 30 minutes à la radioprotection des patients au profit des nouveaux internes
arrivant à l’HIA Legouest.

Article 2 : Evaluation de l’intérêt de l’acquisition abdominopelvienne sans injection lors

de la réalisation d’un scanner corps entier chez un patient suspect de polytraumatisme.

La suppression de la série sans injection lors de la réalisation d’un scanner

abdominopelvien est un moyen simple pour réduire la dose globale du scanner. Dans notre
étude portant sur l’intérêt de la série abdominopelvienne sans injection durant l’acquisition

79
d’un scanner corps entier pour bilan de polytraumatisme, la série sans injection thoraco-
abdominopelvienne était à l’origine de 20 % de la dose totale délivrée au cours du scanner.
Pourtant, l’intérêt diagnostique de l’acquisition abdominopelvienne sans injection s’est avéré
faible et vraisemblablement sans impact sur la prise en charge du patient. De ce fait, sa
suppression permet de réduire la dose globale du patient. Bien que nous n’ayons pas analysé
l’intérêt de la série sans injection au niveau du thorax, nous pensons aussi que son impact est
limité et qu’il est aussi possible de supprimer cette série. Une autre limite de notre étude était
l’absence d’évaluation de l’impact thérapeutique ou du pronostique liée aux lésions
traumatiques analysées. Les résultats de notre étude ont toutefois été confirmés par une étude
plus récente qui retrouvait aussi l’absence d’amélioration de la performance diagnostique de
la série sans injection abdominopelvienne lors de la réalisation d’un scanner corps entier pour
bilan de polytraumatisme [20].

Article 3 : Optimisation de la longueur d’acquisition des scanners réalisés pour colique

néphrétique : proposition d’une nouvelle méthode pour le placement de la limite
supérieure de l’acquisition.

La réduction de la couverture d’acquisition est aussi un moyen simple, rapide et efficace

pour réduire la dose. La couverture d’un scanner doit être centrée sur la zone d’intérêt. Dans
notre étude, nous avons proposé un nouveau repère pour placer le bord supérieur de
l’acquisition d’un scanner réalisé pour suspicion de colique néphrétique. Ce repère est
constitué par le point d’intersection entre le bord antérieur des corps vertébraux et la coupole
diaphragmatique gauche sur le scout de profil. Avec notre repère, nous avons pu réduire de
18,2 % la longueur d’acquisition contre 15,1 % pour la méthode 2 utilisant le bord inférieur
de la vertèbre T10 et 20,5 % pour la méthode 1 utilisant les contours des reins. Parallèlement,
la proportion de reins coupés était significativement supérieure pour la méthode 1 avec 6,7 %
des reins coupés, ce qui apparaît inacceptable en pratique clinique courante. Par contre, il n’y
avait pas de différence significative en termes de proportion de reins coupés entre les
méthodes 2 et 3 avec respectivement 0,7 et 1,4 %. Même si notre méthode a montré qu’elle
coupe un peu plus de rein que la méthode de Corwin MT et al. [46], la distance de rein coupé
reste faible, inférieure à 1 cm, ce qui semble acceptable en pratique clinique courante. Nous
pensons donc que la nouvelle méthode que nous proposons apparaît comme un bon
compromis en permettant une réduction de la longueur d’acquisition plus importante que la

80
méthode 2 mais en coupant moins de rein qu’avec la méthode 1. Dans notre étude, compte
tenu de l’utilisation d’un protocole de scanner avec la modulation automatique du mA, il
n’était pas possible de calculer directement le pourcentage de réduction de la dose
d’irradiation à partir de la réduction de la longueur d’acquisition. Toutefois, la réduction de la
dose est vraisemblablement très proche de la réduction de la longueur d’acquisition. Dans une
étude récente portant sur la réduction de la longueur d’acquisition pour des scanners réalisés
pour suspicion d’appendicite aiguë, Corwin MT et al. ont retrouvé une réduction de la
longueur d’acquisition de 24 % pour une réduction de la dose de 23 % [47]. Par ailleurs, outre
la réduction de la dose globale du scanner par une réduction de la couverture d’acquisition, ce
centrage permet aussi d’éviter au faisceau de rayons X de traverser des organes radiosensibles
comme les gonades chez l’homme et les seins chez la femme. De plus, à cause du phénomène
d’irradiation pré- et post-hélice, qui peut s’étendre sur 2 à 3 cm de part et d’autre de l’hélice,
même si ces organes radiosensibles ne sont pas compris dans les images acquises, ils peuvent
être traversés par le faisceau de rayons X s’ils sont situés à proximité des limites du champ
d’acquisition.

81
82
 CHAPITRE 3 : INFLUENCE DES FACTEURS TECHNIQUES

Ce chapitre est composé de quatre articles :

1- Gervaise A, Louis M, Batch T, Loeuille D, Noel A, Guillemin F, Blum A. Réduction

de dose dans l’exploration du rachis lombaire grâce au scanner 320-détecteurs : étude
initiale. J Radiol 2010; 91: 779-85.

2- Gervaise A, Osemont B, Lecocq S, Micard E, Noel A, Felblinger J, Blum A. CT

image quality improvement using adaptive iterative dose reduction with wide-volume
acquisition on 320-detector CT. Eur Radiol 2012; 22: 295-301.

3- Gervaise A, Osemont B, Louis M, Lecocq S, Teixeira P, Blum A. Standard dose

versus low-dose abdominal and pelvic CT: comparison between filtered back
projection versus adaptive iterative dose reduction 3D. Diagn Interv Imaging 2014;
95: 47-53.

4- Gervaise A, Naulet P, Beuret F, Henry C, Pernin M, Portron Y, Lapierre-Combes M.

Low-dose CT with automatic tube current modulation, adaptive statistical iterative
reconstruction, and low tube voltage for the diagnosis of renal colic: impact of body
mass index. Am J Roentgenol 2014; 202: 553-60.

83
84
 Les facteurs techniques ont bénéficié de nombreuses innovations technologiques ces
dernières années. Nous proposons d’illustrer l’intérêt de plusieurs de ces facteurs techniques
dans une démarche d’optimisation et de réduction de la dose d’irradiation au scanner : le
mode d’acquisition volumique, les reconstructions itératives et la modulation automatique du
milliampérage.

Article 1 : Réduction de dose dans l’exploration du rachis lombaire grâce au scanner

320-détecteurs : étude initiale.

Le mode d’acquisition est l’un des facteurs techniques qui influence la dose et qui est
accessible au moment de l’acquisition. L’apparition des scanners à large système de détection,
comme le scanner 320-détecteurs, permet d’avoir une couverture d’acquisition allant jusqu’à
16 cm en une seule rotation du tube. Pour des scanners avec une faible couverture
d’acquisition, tels que le cœur, le cerveau ou encore les articulations périphériques, il devient
possible de faire des acquisitions volumiques en mode séquentiel et non hélicoïdal. De même,
en juxtaposant deux volumes d’acquisition (= mode Wide-volume ou Stitching mode) il est
aussi possible de faire une acquisition volumique pour des scanners de plus grande longueur
d’acquisition.

Le but de notre étude était de comparer la dose délivrée et la qualité d’image entre un
scanner lombaire réalisé en mode hélicoïdal et en mode Wide-volume grâce au scanner 320-
détecteurs.

Il s’agissait d’une étude monocentrique prospective incluant 20 patients adressés au

service d’imagerie Guilloz du CHRU de Nancy pour la réalisation d’un scanner lombaire. Les
dix premiers patients ont bénéficiés d’un scanner en mode Wide-volume 320-détecteurs et les
dix suivants en mode hélicoïdal 64-détecteurs. Les longueurs d’acquisition, les doses
délivrées ainsi qu’une évaluation quantitative et qualitative de la qualité d’image ont été
comparées entre les deux groupes.

Les résultats montraient une réduction significative de la dose délivrée de 35 % en faveur

du mode d’acquisition volumique versus hélicoïdal (produit dose longueur de respectivement
970 [Link] versus 1503 [Link]). Par contre, il n’y avait pas de différence statistiquement
significative de qualité d’image ou de longueur d’acquisition entre les deux groupes.

85
 Article 2 : Amélioration de la qualité d’image scanographique en utilisant les
reconstructions itératives Adaptive Iterative Dose Reduction avec une acquisition wide-
volume sur un scanner 320-détecteurs.

Malgré le développement de plusieurs innovations technologiques (modulation

automatique du milliampérage, collimation active), la réduction de la dose des scanners reste
limitée par l’utilisation des reconstructions standard en FBP car celles-ci entraînent une
augmentation notable du bruit de l’image en cas de réduction trop importante de la dose [41].
L’apparition récente des reconstructions itératives a permis une réduction significative du
bruit de l’image par rapport aux reconstructions standard FBP [42].

Le but de cette étude était d’évaluer l’impact sur fantôme et sur patient des reconstructions
itératives AIDR sur la qualité d’image et sur la dose.

L’étude sur fantôme était réalisée à partir d’un fantôme Catphan 500® en faisant des
acquisitions volumiques avec le scanner 320-détecteurs avec un milliampérage allant de 25 à
550 mAs (milliampère x secondes). Les images étaient reconstruites en FBP et en AIDR. Le
bruit de l’image, le RCB, le RSB et la résolution spatiale des images ont été comparés entre
les images FBP et AIDR. Les reconstructions itératives AIDR ont ensuite été testées sur 15
scanners lombaires au cours d’une étude prospective monocentrique sur patient. Les images
étaient reconstruites en FBP et AIDR. Le bruit de l’image et le RSB ont été comparés entre
les deux séries d’image.

Sur le fantôme, les résultats montraient une réduction significative du bruit de l’image de
40 % en faveur des reconstructions itératives AIDR par rapport aux images FBP. Il existait
aussi une amélioration significative du RSB et du RCB avec les reconstructions AIDR. Par
contre, il n’y avait pas de différence significative de résolution spatiale entre les deux types
d’images. Au niveau des scanners lombaires, il existait une amélioration significative de
l’évaluation quantitative et qualitative de la qualité d’image sur les images AIDR versus FBP.

Article 3 : Réduction de la dose des scanners abdominopelviens grâce aux reconstructions

itératives AIDR 3D.

Tandis que les reconstructions itératives AIDR n’étaient disponibles que pour une
acquisition volumique sur le scanner 320-détecteurs et avec une reconstruction rétrospective

86
des images, l’introduction des nouvelles reconstructions itératives AIDR 3D a permis leur
utilisation en mode d’acquisition hélicoïdal prospectif. Il a ainsi été possible de les utiliser en
pratique clinique courante.

Le but de cette étude était de comparer la dose et la qualité d’image d’un scanner
abdomino-pelvien en dose normale avec les reconstructions standard FBP versus en basse-
dose avec les reconstructions itératives AIDR 3D et les reconstructions FBP.

Il s’agissait d’une étude rétrospective incluant 21 patients ayant eu un scanner abdomino-

pelvien au temps portal avant (série FBP dose standard) et après l’implantation des
reconstructions itératives AIDR 3D (série AIDR 3D basse-dose et FBP basse-dose). La
longueur d’acquisition, la dose délivrée ainsi qu’une évaluation de la qualité d’image ont été
comparées entre les trois séries d’images à partir d’un test des rangs signés de Wilcoxon.

Les résultats montraient une réduction significative de la dose des scanners abdomino-
pelviens de 49,5 % après l’implantation des reconstructions itératives AIDR 3D (moyenne des
PDL de 451 [Link] contre 892 [Link], p < 0,001). Aucune différence significative n’était
observée entre les séries FBP dose standard et AIDR 3D basse-dose concernant l’évaluation
de la qualité d’image (score de qualité d’image de respectivement 4,6 ± 0,6 versus 4,4 ±
0,6 avec p = 0,147).

Article 4 : Scanner basse dose avec la modulation automatique du milliampérage, les

reconstructions Adaptive Statistical Iterative Reduction et un faible kilovoltage pour le
diagnostic des coliques néphrétiques : impact de l’indice de masse corporelle.

Le mA correspond à l’intensité du courant du tube radiogène et représente la quantité de

photons produite au sein du faisceau de rayons X. Le mA est proportionnel à la dose délivrée
et inversement proportionnel au carré du bruit de l’image [30]. L’optimisation du mA doit
prendre en compte le morphotype du patient. En effet, la quantité de photons X nécessaire
pour garder une qualité de l’image constante est variable en fonction de l’épaisseur du patient.
Pour mieux adapter le mA au morphotype des patients, des techniques de modulation
automatique du mA ont été développés au début des années 2000 [31]. Grâce à ces techniques,
le mA est automatiquement adapté au morphotype du patient et à la zone anatomique à
explorer et le bruit de l’image reste constant sur l’ensemble des images de l’acquisition. Par

87
ailleurs, ces techniques ont permis de réduire les doses délivrées. Leur impact sur la
performance diagnostique en fonction du morphotype des patients reste encore à étudier.

L’objectif de notre étude était d’évaluer l’impact du morphotype des patients sur la dose,
la qualité d’image et la performance diagnostique de notre protocole de scanner basse dose
sans injection réalisé avec un faible kilovoltage, la modulation automatique du mA et ASIR
chez des patients avec une suspicion clinique de colique néphrétique.

Il s’agissait d’une étude rétrospective monocentrique incluant l’ensemble des patients

adressés dans le service d’imagerie de l’HIA Legouest durant l’année 2012 pour la recherche
d’une colique néphrétique et ayant eu un scanner sans injection selon notre protocole basse
dose avec les reconstructions itératives ASIR, un faible kilovoltage et la modulation
automatique du mA. Les images du scanner basse dose ont été analysées par deux radiologues
et une interne en radiologie qui ont évalué la présence d’une colique néphrétique, la confiance
dans le diagnostic et la qualité d’image. Ces résultats ainsi que les doses délivrées ont été
comparés chez des patients avec des catégories d’Indice de Masse Corporelle (IMC)
différents.

Les résultats montraient qu’il n’y avait pas de différence statistiquement significative de
performance diagnostique entre le groupe de patients avec un IMC < 25 kg/m² et les patients
avec un IMC > 25 kg/m². Par contre, les scores de qualité d’image et de confiance dans le
diagnostic étaient significativement meilleurs chez les patients avec un IMC > 25 kg/m² par
rapport aux patients avec un IMC < 25 kg/m² (respectivement 3,7 versus 3,4 avec p < 0,001 et
2,8 versus 2,5 avec p < 0,001). La dose efficace moyenne des scanners était également
supérieure pour les patients avec un IMC > 25 kg/m² comparativement aux patients avec un
IMC < 25 kg/m² (3,7 mSv versus 2,4 mSv).

88
 Chapitre 3

Article 1 : Réduction de dose dans l’exploration du rachis lombaire grâce au scanner 320-
détecteurs : étude initiale.

Gervaise A, Louis M, Batch T, Loeuille D, Noel A, Guillemin F, Blum A. Réduction de dose dans l’exploration
du rachis lombaire grâce au scanner 320-détecteurs : étude initiale. J Radiol 2010;91: 779-85.

89
90
 J Radiol 2010;91:779-85
© Éditions Françaises de Radiologie, Paris, 2010
Édité par Elsevier Masson SAS. Tous droits réservés article original ostéoarticulaire

Réduction de dose dans l’exploration du rachis

lombaire grâce au scanner 320-détecteurs :
étude initiale
A Gervaise (1), M Louis (1), T Batch (1), D Loeuille (2), A Noel (3), F Guillemin (4) et A Blum (1)

Abstract Résumé
Dose reduction at CT of the lumbar spine using a 320-detector row Objectif. Comparer la dose délivrée et la qualité d’image entre un
scanner: initial results scanner lombaire réalisé en mode hélicoïdal et en mode Wide volume
J Radiol 2010;91:779-85 grâce au scanner 320-détecteurs.
Patients et méthodes. Il s’agit d’une étude monocentrique prospective
Purpose. To compare radiation dose and image quality for CT of the incluant 20 patients consécutifs répartis en deux groupes. Les
lumbar spine between helical CT and wide volume mode scanning 20 patients ont bénéficié d’un scanner lombaire sur le scanner 320-
with a 320-detector row CT. détecteurs (Aquilion One, Toshiba). Les scanners lombaires du
Patients and methods. Monocenter prospective study on 20 consecutive groupe 1 ont été réalisés en mode Wide volume 320-détecteurs et
patients divided into two groups. All 20 patients underwent lumbar ceux du groupe 2 en mode hélicoïdal 64-détecteurs. La longueur
spine CT on the 320-detector row scanner (Aquilion One, Toshiba). d’acquisition, la dose délivrée correspondant au PDL. e (Produit
The CT examinations for group 1 were performed using the wide Dose Longueur. étendu) ainsi qu’une évaluation quantitative et
volume mode with 320 detector rows while the CT examinations for qualitative de l’image ont été comparées entre les deux groupes.
group 2 were performed using a 64-detector row helical CT mode. The Résultats. Les moyennes de longueur d’acquisition étaient compa-
acquisition length and delivered dose corresponding to the DLPe rables entre les deux groupes. Il existait une réduction significative de
(extended dose length product) as well as qualitative and quantitative la dose délivrée d’environ 35 % (moyenne des PDL. e de 970 mGy. cm
image quality were compared between both groups. pour le groupe 1 contre 1 503 mGy. cm pour le groupe 2, p < 0,028)
Results. The mean acquisition length was comparable between both avec le mode Wide volume 320-détecteurs par rapport au mode
groups. There was a significant dose reduction of about 35% for group hélicoïdal 64-détecteurs. Aucune différence significative n’était
1 compared to group 2 (mean DLPe of 970 [Link] for group 1 observée entre les deux techniques concernant l’évaluation de la
compared to 1503 [Link] for group 2, p<0.028) when using the wide qualité de l’image.
volume mode acquisition at 320-detector row CT compared to the 64- Conclusion. L’acquisition d’un scanner lombaire en mode Wide
detector row helical CT mode. No significant difference was noted for volume sur le scanner 320-détecteurs permet de réduire significati-
image quality between both groups. vement la dose délivrée au patient par rapport à l’acquisition
Conclusion. The acquisition of lumbar CT using the wide volume hélicoïdale 64-détecteurs, tout en préservant une qualité d’image
mode at 320-detector row CT allows significant dose reduction to équivalente.
patients compared to the 64-detector row helical CT mode while
preserving image quality.

Key words: CT. 320-detector row. Dose. Reduction. Lumbar spine. Mots-clés : Scanner. 320-détecteurs. Dose. Réduction. Rachis
lombaire.

L
es lombo-radiculalgies sont un problè- nance magnétique sont équivalentes pour la tion constante du radiologue, en accord avec
me majeur de santé publique (1). En recherche d’un conflit disco-radiculaire (3), le principe de précaution ALARA (As Low
France, plus de 70 % des travailleurs cette dernière souffre d’un manque de dis- As Reasonably Achievable). Cela est
ont déjà présenté un épisode de lombalgie et ponibilité et d’un coût plus élevé. Son utili- d’autant plus vrai dans l’exploration scano-
un tiers de ces 70 % a déjà eu un arrêt de tra- sation est également limitée par certaines graphique du rachis lombaire qui concerne
vail en rapport à cette symptomatologie (2). contre-indications (obésité, claustrophobie, principalement des patients jeunes (18, 19)
De nombreuses explorations radiologiques pacemaker) et par une reproductibilité inte- et susceptibles de recourir à des examens
sont réalisées dans le cadre du bilan de ces robservateur modérée voire faible, notam- scanographiques répétés (20).
douleurs. Tandis que la sensibilité et la spé- ment pour l’étude des articulations intera- C’est dans ce contexte que le scanner 320-
cificité du scanner et de l’imagerie par réso- pophysaires et des sténoses canalaires ou détecteurs semble pouvoir réduire la dose
foraminales (4-7). Cela explique en partie délivrée au patient par rapport aux scanners
que le scanner lombaire reste l’un des exa- hélicoïdaux classiques. Ce nouveau type de
(1) Service d’Imagerie Guilloz, Hôpital Central, CHU
Nancy, 29 avenue du Maréchal de Lattre de Tassigny,
mens scanographiques les plus pratiqués en scanner bénéficie d’un large système de dé-
54035 Nancy cedex. (2) Service de Rhumatologie, Hôpi- France (8). Pourtant, le scanner est une tech- tection capable d’obtenir en une seule rota-
tal de Brabois, CHU Nancy, allée du Morvan, 54511
Vandœuvre-lès-Nancy. (3) Unité de radiophysique mé- nique irradiante. Le lien entre un éventuel tion de 0,35 s un volume de 160 mm dans
dicale, CRAN UMR 7039 Nancy Université-CNRS, risque cancérigène et l’exposition à des fai- l’axe z. Cette caractéristique technique uni-
Centre Alexis Vautrin, avenue de Bourgogne, 54511
Vandœuvre-lès-Nancy. (4) Inserm CIC-EC, épidémio- bles doses de rayons X reste très controversé que permet de réduire considérablement le
logie et évaluation cliniques, CHU de Nancy, 54511 (9-11) et fait l’objet de nombreuses publica- temps d’acquisition, donc les artefacts de
Vandœuvre-lès-Nancy.
Correspondance : A Gervaise tions (12-17). La réduction de la dose déli- mouvements, tout en étant capable de
E-mail : [Link]@[Link] vrée au patient doit donc être une préoccupa- couvrir des organes entiers. Ce mode

© 2016 Elsevier Masson SAS. Tous droits réservés. - Document téléchargé le 15/03/2016
 780 Réduction de dose dans l’exploration du rachis lombaire A Gervaise et al.
grâce au scanner 320-détecteurs : étude initiale

« volumique » est ainsi largement utilisé Technique d’acquisition de dosimétrie classique compte tenu de la
dans le domaine de l’imagerie cardiaque largeur importante du système de détec-
(21, 22) et vasculaire (23), pour l’étude de Un scanner lombaire sans injection a été tion du scanner 320-détecteurs). Les va-
la perfusion cérébrale (24) ou d’organes réalisé chez tous les patients sur le même leurs de PDL.e relevées dans le groupe
abdominaux (25) et a également trouvé scanner 320-détecteurs (Aquilion ONE, Wide volume étaient en accord avec
ses indications en imagerie pédiatrique Toshiba Medical Systems, Otawara, Japon). la formule décrite pour les scanners volu-
(26). Dans toutes ces applications, il a été Les patients étaient positionnés en décubi- miques dynamiques (28) : PDL.e
montré une réduction significative de la tus dorsal, pieds les premiers. L’acquisition = CTDIvol.e × BWnom avec CTDIvol.e
dose délivrée pouvant atteindre 75 % (21). débutait par la réalisation d’un topogramme = CTDIw.e × R où le CTDIvol.e est l’in-
La juxtaposition de plusieurs volumes de face et de profil. Celui-ci permettait de dex de dose scanographique volumique
permet également l’exploration de zones repérer la zone à explorer et d’utiliser la mo- étendu (Computed Tomographic Dose
anatomiques plus étendues. Ce nouveau dulation automatique de la dose dans les Index volume. extended), BWnom est la
mode d’acquisition, qui est l’équivalent 3 plans de l’espace (x, y et z) afin d’adapter largeur de faisceau nominal (Beam
d’un mode « incrémental volumique », les paramètres d’acquisition à la corpulence Width), le CTDIw.e est le CTDI pondéré
est appelé mode Wide volume ou « Stit- de chaque patient pour obtenir un niveau étendu et R est le nombre de rotations.
ching mode » et permet de couvrir le ra- de qualité d’image déterminé.
chis lombaire grâce à la juxtaposition de Dans le groupe 1, le scanner était réalisé en Évaluation de la qualité d’image
deux volumes. Dans leur protocole de bi- mode Wide volume 320-détecteurs grâce à
2 volumes couvrant T12 à S2 dont la sépa- La qualité d’image a été évaluée de ma-
lan de douleur thoracique, Hein et al. (27)
ration se faisait à hauteur des crêtes ilia- nière quantitative par la mesure du bruit
ont montré que la triple acquisition tho-
ques (135 kV, modulation automatique de l’image. Celui-ci était estimé à partir
racique et cardiaque en mode volumique
des milliAmpère (mA) dans les 3 plans de la déviation standard (en Unité
et Wide volume 320-détecteurs était ac-
(maximum 500 mA et minimum 100 mA) Hounsfield) d’une région d’intérêt (ROI)
compagnée d’une réduction de dose d’en-
avec pour référence un indice de bruit (IB) de 1 cm2 placée dans le muscle psoas
viron 55 %. À notre connaissance, aucune
à 7,5 pour une coupe de 5 mm avec un fil- droit, sur une coupe axiale de 1 mm en fe-
étude n’a encore évalué cette réduction
tre standard mou, temps de rotation (TR) à nêtre parenchymateuse, à hauteur des pé-
pour l’acquisition d’un scanner lombaire
0,75 s, 320 × 0,5 mm avec reconstruction en dicules de L5. Les ROI ont été placées
en mode Wide volume. Le scanner 320-
coupe axiale de 1 mm tout les 1 mm, filtre chez les 20 patients sur une console IM-
détecteurs possède un autre avantage :
de reconstruction FC 08). PAX par le même radiologue.
grâce à la désactivation d’un certain nombre
Dans le groupe 2, le scanner était réalisé de L’évaluation qualitative a été réalisée sur
de détecteurs, il peut fonctionner comme un
T12 à S2 en mode hélicoïdal 64-détecteurs les coupes axiales natives de 1 mm en fenê-
scanner hélicoïdal (avec 64-canaux de détec-
(135 kV, modulation automatique des mA tre parenchymateuse à partir d’une échelle
tion et prochainement 160). Il est ainsi possi-
dans les 3 plans (maximum 500 mA et mi- de score allant de 0 à 4 (0 = qualité d’image
ble de comparer sur la même machine la
nimum 100 mA) avec pour référence un médiocre ne permettant pas une interpré-
dose délivrée pour l’acquisition d’un scan-
IB à 6 pour une coupe de 5 mm avec un tation ; 1 = mauvaise qualité de l’image in-
ner lombaire en mode Wide volume 320-
filtre standard mou, pitch à 0,641, TR à terférant avec la qualité diagnostique de
détecteurs versus hélicoïdal 64-détecteurs.
0,75 s, 64 × 0,5 mm avec reconstruction en l’examen ; 2 = image de qualité moyenne ;
L’objectif de cette étude est donc de me-
coupe axiale de 1 mm tout les 1 mm, filtre 3 = bonne qualité de l’image ; 4 = excellente
surer la réduction de la dose délivrée au
de reconstruction FC 08). qualité de l’image). Cette évaluation a été
patient lors de la réalisation d’un scanner
réalisée indépendamment par 3 radiolo-
lombaire entre une acquisition en mode Après avoir été automatiquement en-
gues, après une séance de lecture commu-
Wide volume 320-détecteurs et hélicoïdal voyés dans le PACS, les images et le rap-
ne, sur des consoles IMPAX, après anony-
64-détecteurs tout en comparant la quali- port de dose du scanner étaient accessibles
misation et en l’absence d’affichage des
té d’image afin de s’assurer qu’une réduc- à partir d’une station de travail IMPAX
paramètres d’acquisition.
tion de dose ne s’accompagne pas d’une V5 (Agfa, Ridgefield Park, NJ, USA).
altération de la qualité d’image.
Évaluation de la longueur Analyse statistique
d’acquisition Compte tenu du faible effectif de patients
Patients et méthodes inclus, un test des rangs signés de Wil-
La longueur d’acquisition était exprimée coxon était utilisé pour comparer les va-
Population étudiée en cm et correspondait à la différence de leurs de longueur d’acquisition, de PDL.
position entre la première et la dernière e et de qualité d’image entre les deux
Il s’agit d’une étude monocentrique pros- image de l’acquisition. groupes. Une valeur de p inférieure à 0,05
pective incluant 20 patients consécutifs
était considérée comme significative.
entre février et mars 2009. Les 10 premiers
patients ont bénéficié d’un scanner lombaire
Évaluation de la dose délivrée
en mode Wide volume 320-détecteurs La dose délivrée était directement fournie
(groupe 1) et les 10 suivants en mode héli- par le rapport d’examen. Elle correspon- Résultats
coïdal 64-détecteurs (groupe 2). Seuls les dait au PDL.e (Produit Dose Longueur.
patients de plus de 45 ans ont été inclus. étendu) exprimé en [Link] (le terme La moyenne d’âge et l’IMC moyen du
Pour chaque patient, l’âge et l’indice de « étendu » est lié à la nécessité d’extrapo- groupe 1 étaient de respectivement
masse corporelle (IMC) étaient notés. ler les valeurs de PDL à partir des relevés 65,1 ans (de 53 à 88 ans) et 28,3 kg/m2 (de

J Radiol 2010;91

© 2016 Elsevier Masson SAS. Tous droits réservés. - Document téléchargé le 15/03/2016
 A Gervaise et al. Réduction de dose dans l’exploration du rachis lombaire 781
grâce au scanner 320-détecteurs : étude initiale

19,1 à 46,5 kg/m2) contre 61,4 ans (de 46 à Wide volume 320-détecteurs permet de gueur supplémentaire irradiée de 15 mm
74 ans) et 27,8 kg/m2 (de 20,2 à 33,8 kg/m2) réduire de façon significative la dose déli- en début et fin d’hélice pour une même
pour le groupe 2. vrée au patient sans altérer la qualité longueur d’acquisition de 160 mm. Dans
Les longueurs d’acquisition étaient d’image. Cette réduction peut s’expliquer notre étude, l’overranging en mode héli-
comparables entre les 2 groupes. Il exis- par trois principales caractéristiques tech- coïdal a été estimé à environ 15 % de la
tait une différence significative entre niques. dose totale. L’abandon de l’hélice grâce
les PDL.e du groupe 1 et du groupe 2 Tout d’abord, en mode hélicoïdal, pour au mode Wide volume supprime ce phé-
(tableau I) avec une réduction de 35 % de s’assurer de l’entièreté des premières et nomène d’overranging et permet de ré-
la dose délivrée en mode Wide volume dernières images de l’acquisition, il est duire d’autant la dose délivrée au patient.
par rapport au mode hélicoïdal. Aucune nécessaire de faire un tour supplémentai- Chez les jeunes femmes, cette suppres-
différence significative n’était mise en re à chaque extrémité de la zone explorée. sion est bénéfique puisque l’exposition
évidence concernant l’évaluation qualitati- Cette exposition « pré et post-hélice », post-hélice d’un scanner lombaire concer-
ve ou quantitative de la qualité d’image également appelée overranging ou z ne des organes radiosensibles que sont les
(tableau II). overscanning, varie proportionnellement ovaires. L’installation prochaine sur
en fonction du pitch et du nombre de dé- l’Aquilion One d’une collimation active
tecteurs et est inversement proportionnel- par bouclier permettra toutefois de rédui-
re de manière importante la part de
le à la longueur d’exploration (29, 30). Ce
Discussion l’overranging en mode hélicoïdal (sans
phénomène est visualisé sur la figure 1.
toutefois pouvoir égaler la réduction in-
Notre étude confirme que la réalisation L’utilisation d’un film radiochromique
duite par la suppression de l’hélice).
d’un scanner lombaire grâce au mode permet de mettre en évidence une lon-
Deuxièmement, la réduction de dose s’ex-
plique par une moindre importance de
l’effet d’overbeaming. Ce dernier corres-
Tableau I pond au phénomène de pénombre : afin
Comparaison des longueurs d’acquisition et des doses délivrées entre les modes Wide de couvrir l’ensemble des détecteurs avec
volume et hélicoïdal. un rayonnement d’intensité égale, le fais-
Mode Wide volume Mode hélicoïdal Valeur de p ceau de rayons X doit déborder du champ
Longueur des détecteurs (fig. 2). Cela entraîne une
26,23 ± 1,75 26,99 ± 1,33 p > 0,578
d’acquisition (cm) exposition supplémentaire qui ne contri-
PDL. e (mGy. cm) 970,9 ± 359,8 1 503,2 ± 324,1 p < 0,028 bue pas à la formation de l’image mais
Les données correspondent aux moyennes des mesures ± la déviation standard. participe à la dose délivrée au patient (31).
L’importance de l’overbeaming diminue
proportionnellement avec la largeur du
Tableau II système de détection (32) et est donc pro-
Évaluation quantitative et qualitative de la qualité de l’image des scanners lombaires en portionnellement moins importante pour
mode Wide volume et hélicoïdal. un scanner à large système de détection
Mode Wide volume Mode Valeur de p (33). L’efficience de dose (encore appelée
hélicoïdal z efficiency) informe l’utilisateur sur la
Évaluation quantitative proportion de rayonnement qui ne
Déviation standard contribue pas à la formation de l’image.
24,00 ± 6,84 20,90 ± 4,77 p > 0.359
de la ROI (UH) Cette valeur dépend principalement de
Evaluation qualitative l’épaisseur de coupe, et dans une moindre
Lecteur 1 3,4 ± 0,52 3,6 ± 0,52 p > 0,375 mesure, de l’effet d’overbeaming. Cepen-
Lecteur 2 3,3 ± 0,67 3,6 ± 0,52 p > 0,437 dant, l’épaisseur de coupe étant constante
Lecteur 3 3,4 ± 0,52 3,7 ± 0,48 p > 0,312 entre nos deux modes d’acquisition, l’effi-
Les données correspondent aux moyennes des mesures ± la déviation standard. cience de dose permet d’évaluer l’effet
d’overbeaming. Sa valeur est donc d’autant
plus faible que l’effet d’overbeaming, et
donc l’exposition non contributive à
l’image, augmente. Ainsi, tandis que
pour la réalisation d’un volume de
160 mm, l’efficience de dose est d’environ
93 % en mode volumique 320-détecteurs,
elle ne dépasse pas 85 % pour le mode hé-
licoïdal 64-détecteurs. Par ailleurs, l’effet
d’overbeaming se répète à chaque rota-
Fig. 1 : Mise en évidence du phénomène d’overranging à partir d’un papier radiochromique
(en présence de rayons X, la zone irradiée s’assombrit). Deux acquisitions de même tion du tube. La diminution du nombre
longueur sont réalisées en mode hélicoïdal (papier du haut) et volumique (papier du de rotations en mode Wide volume (seu-
bas). Pour la même longueur d’acquisition, la zone irradiée est plus importante en
mode hélicoïdal : cette exposition « pré et post-hélice » correspond à l’overranging
lement 2 en mode Wide volume contre
(doubles flèches). environ 13 tours d’hélice pour couvrir le

J Radiol 2010;91

© 2016 Elsevier Masson SAS. Tous droits réservés. - Document téléchargé le 15/03/2016
 782 Réduction de dose dans l’exploration du rachis lombaire A Gervaise et al.
grâce au scanner 320-détecteurs : étude initiale

de reconstruction adapté (ConeXact pour

l’Aquilion One) permettant d’éliminer
ces artefacts. Cet algorithme de recons-
truction est toutefois très performant
puisqu’il est possible de reconstruire
640 coupes à partir des 320 coupes natives
d’un volume de 160 mm. Cet algorithme
de reconstruction « double-coupe » per-
met donc de disposer de reconstructions
chevauchées de 0,5 mm tout les 0,25 mm
à l’origine d’une diminution des effets de
cône et des artefacts de volume partiel.
Par contre, l’utilisation du mode Wide
volume nécessite un chevauchement par-
Fig. 2 : Profil de dose d’un volume de 160 mm : le faisceau déborde du champ des détec-
tiel des volumes afin de s’assurer qu’il n’y
teurs. Ce phénomène correspond à l’overbeaming. Le rapport entre la largeur de la a pas de données manquantes. Ce chevau-
zone explorée (160 mm) et la largeur de la zone irradiée (171,9 mm) correspond à chement entraîne une exposition plus im-
l’efficience de dose (93 % dans cet exemple). Noter également la forme asymétrique
du profil de dose due à l’effet talon de l’anode. portante à la jonction des deux volumes.
Le raccourcissement de cette zone de che-
vauchement par l’amélioration de l’algo-
rithme de reconstruction est une des solu-
rachis lombaire) contribue aussi à réduire tions sont toutes acquises au même niveau tions pour permettre de réduire cette
l’importance de l’exposition due à l’over- anatomique pour reconstruire une coupe. surexposition.
beaming en mode Wide volume. Cela procure une amélioration du RSB Le principe de la modulation automatique
Enfin, en mode hélicoïdal, le déplace- permettant une réduction des paramètres de la dose est également différent entre les
ment de la table lors de l’acquisition de d’exposition. Par contre, la géométrie deux modes d’acquisition. Tandis qu’en
l’hélice nécessite une interpolation des particulière du faisceau, secondaire à la mode hélicoïdal l’adaptation du milliam-
données à l’origine d’une dégradation du largeur importante du système de détec- pérage se fait tout au long de l’hélice et
rapport signal à bruit (RSB) et du profil tion en mode Wide volume 320-détec- permet de réduire la dose délivrée jusqu’à
de coupe (34). En mode Wide volume, teurs (angle d’ouverture du faisceau de 33 % (35), une seule valeur de milliampé-
comme en mode incrémental convention- 15,2° contre 3,05° pour le mode hélicoïdal rage est disponible par volume. La modu-
nel, il n’y a pas d’interpolation des données 64-détecteurs), engendre des effets de cône lation automatique de dose pour le scan-
qui sont cohérentes puisque les projec- qui nécessite l’utilisation d’un algorithme ner lombaire en mode Wide volume ne
bénéficie donc que de deux valeurs de
milliampérage (une pour chacun des
deux volumes). Pour adapter cette modu-
lation binaire à l’acquisition d’un scanner
lombaire, nous avons placé les 2 volumes
de part et d’autre des crêtes iliaques
(fig. 3). Ce placement suit effectivement
l’évolution anatomique puisque l’épais-
seur du bassin entraîne une atténuation
plus importante du faisceau de rayons X
nécessitant une augmentation du mil-
liampérage pour garder une qualité
d’image optimale. Malgré cette adapta-
tion, on peut supposer que la modulation
automatique de la dose est moins efficace
en mode Wide volume par rapport au
mode hélicoïdal. La comparaison des
coefficients de corrélation entre IMC et
dose délivrée dans chaque groupe serait
intéressante pour estimer l’efficacité de
a b celle-ci. Malheureusement, compte tenu
Fig. 3 : Modulation du milliampérage en mode hélicoïdal et Wide volume. du faible effectif de notre étude, nous
a En mode hélicoïdal, le milliampérage varie tout au long de l’hélice. Compte tenu de
l’épaisseur du bassin, on note que le milliampérage augmente à partir des crêtes
n’avons pas réalisé ce calcul.
iliaques. De plus, cette modulation est à l’origine
b En mode Wide volume, chaque volume possède sa propre valeur de milliampérage. d’une discrète différence de densité à l’in-
La position des 2 volumes dont la jonction se fait à hauteur des crêtes iliaques per-
met de rapprocher la courbe de milliampérage du mode Wide volume avec celle du terface des deux volumes due à la diffé-
mode hélicoïdal. rence de milliampérage (fig. 4). A celle-ci

J Radiol 2010;91

© 2016 Elsevier Masson SAS. Tous droits réservés. - Document téléchargé le 15/03/2016
 A Gervaise et al. Réduction de dose dans l’exploration du rachis lombaire 783
grâce au scanner 320-détecteurs : étude initiale

partir de ces chambres d’ionisation. Cela fantômes (43). Ces doses ne sont donc pas
est d’autant plus vrai qu’il faut que la lar- comparables à celles de population dont le
geur de la chambre d’ionisation prenne morphotype n’est pas « standard ». Ainsi,
également en compte l’overbeaming ainsi dans notre étude, seulement 4 patients du
que le rayonnement diffusé. Pour Mori et groupe 1 et 3 du groupe 2 avaient un IMC
al. (38), il faut donc une chambre d’ionisa- compris entre 18 et 25, or l’augmentation
tion d’au moins 300 mm pour permettre de la corpulence nécessite une majoration
de mesurer le CTDIvol lors de l’acquisition des doses délivrées afin de garantir une
d’un volume de 160 mm. Toutefois, plu- qualité d’image équivalente. De plus, les
sieurs études ont montré qu’il était possible longueurs d’acquisition sont souvent dif-
d’extrapoler, grâce à une simulation de férentes. Tandis que la longueur d’explo-
Monte Carlo, la valeur du CTDIvol d’un ration du rachis lombaire n’excède pas
volume de 160 mm à partir des mesures 7 cm dans certaines études (44), notre lon-
de dosimétrie d’une chambre d’ionisation gueur d’acquisition moyenne était de
standard de 100 mm (33, 39). La mesure 26 cm. Notre choix est critiquable mais
du CTDIvol peut donc se faire avec le ma- est lié à la volonté de couvrir systémati-
tériel standard mais la nécessité d’utiliser quement la jonction dorso-lombaire afin
une formule d’extrapolation explique que de permettre la recherche d’éventuels si-
les données de dosimétrie du scanner 320- gnes de spondylarthropathie ou de faire le
détecteurs correspondent au CTDIvol. e bilan d’un canal lombaire étroit. On peut
Fig. 4 : Reconstruction sagittale d’un et au PDL. e (« e » pour « extended » ou d’ailleurs remarquer l’intérêt de l’utilisa-
scanner lombaire en mode Wide
volume : discrète différence de étendu). Enfin, le PDL rend compte de la tion du mode Wide volume : tandis que
densité entre les deux volumes, dose délivrée au cours d’une procédure les doses en mode hélicoïdal sont supé-
due à la différence de milliampé-
rage, à l’origine d’une ligne de
complète (40) et permet d’estimer la dose rieures à la dose moyenne de 19 mSv de
démarcation (têtes de flèches). efficace (41). Le PDL est donc directe- l’étude de Biswas et al (46), l’utilisation du
ment en rapport avec le risque stochasti- mode Wide volume permet de réduire ces
que pour le patient et c’est pourquoi nous doses sous cette moyenne tout en gardant
peut également se rajouter un artefact de avons choisi d’utiliser cette valeur pour une longueur d’exploration importante.
décalage entre les deux volumes contigus. comparer les doses délivrées entre les Notre étude comporte plusieurs limites.
Ce décalage est engendré par les mouve- deux groupes. La première réside dans le faible effectif
ments du patient ou les minimes mouve- Bien qu’il ne s’agisse pas de l’objectif de patients inclus et dans l’absence de
ments de table entre les deux acquisitions. principal de notre étude, ces relevés de randomisation des patients entre les
Bien que cet artefact soit quelque fois visi- dosimétrie sont l’occasion d’évaluer les deux groupes. Ainsi, même si la diffé-
ble sur les reconstructions sagittales ou doses délivrées pour un scanner lombaire rence de dose est significative, une étude
frontales, il est en fait très peu marqué dans notre service. On peut tout de suite randomisée avec un effectif plus grand
pour le scanner lombaire compte tenu de remarquer que le scanner lombaire ne fait est nécessaire pour confirmer l’absence
la courte durée séparant l’acquisition des pas partie des examens soumis aux ni- de différence significative en terme de
deux volumes (1,4 s) et de sa correction veaux de référence diagnostiques (NRD) qualité d’image. Deuxièmement, nos
lors de la reconstruction des images. (42) bien qu’il représente un nombre im- critères d’évaluation de la qualité d’ima-
Une autre particularité du scanner 320- portant d’examens réalisés chaque année. ge sont également discutables. Pour
détecteurs réside dans le calcul des don- Le rapport 2008 de l’IRSN (8), portant l’évaluation quantitative, nous avons op-
nées de dosimétrie. Classiquement, les sur la mise à jour des NRD, propose té pour la mesure du bruit de l’image à
paramètres de dosimétrie sont automati- d’ailleurs que le scanner lombaire figure partir de la déviation standard d’une
quement transmis par le scanner sous la parmi les régions anatomiques concernées ROI, comme cela a déjà été décrit par
forme du CTDIvol et du PDL. Le CTDI- par ce recueil des données dosimétriques. plusieurs auteurs (50, 51). Toutefois, le
vol correspond à la dose moyenne pour Par rapport aux données de la littérature, bruit de l’image n’est pas un critère suf-
une coupe en prenant en compte le pas de on peut également observer que les doses fisant et d’autres critères, plus difficile-
l’hélice (CTDIvol = CTDIw/Pitch) (30). délivrées pour la réalisation d’un scanner ment analysables, participent également
Le CTDIvol est donc une mesure adaptée lombaire sont très variables, allant de 3 à à appréhender la qualité d’image (com-
à l’acquisition hélicoïdale mais perd sa si- 19 mSv (16, 43-48). Ces valeurs corres- me le contraste par exemple ou encore la
gnification lors de l’utilisation d’un scan- pondent aux doses efficaces exprimées en différence de signal rapportée au bruit)
ner volumique (33). Par ailleurs, les milliSievert (mSv). Dans notre étude, les (52). L’évaluation qualitative d’une ima-
chambres d’ionisation, actuellement utili- doses efficaces étaient de 14 mSv pour le ge n’est pas non plus une méthode très
sées et définies par la législation pour faire groupe 1 et 22,5 mSv pour le groupe 2 (la robuste d’autant plus que notre évalua-
les relevés de dosimétrie (36, 37), ont une dose efficace (E) est calculée en utilisant le tion n’a pu se faire qu’à partir des coupes
longueur active de 100 mm. Elles ne sont coefficient de conversion tissulaire (k) de axiales. En effet, sur les reconstructions
donc pas adaptées pour mesurer le CTDI- l’abdomen à 0,015 selon la formule E sagittales ou frontales, les possibles arte-
vol du scanner 320-détecteurs car il est im- = PDL. e × k (49)). Cette disparité tient facts à la jonction des deux volumes ris-
possible d’enregistrer l’intégralité du pro- d’abord au fait que certaines études pren- quaient d’informer le lecteur du groupe
fil de coupe d’un volume de 160 mm à nent en compte les doses délivrées sur des auquel appartenait le patient. Pourtant,

J Radiol 2010;91

© 2016 Elsevier Masson SAS. Tous droits réservés. - Document téléchargé le 15/03/2016
 784 Réduction de dose dans l’exploration du rachis lombaire A Gervaise et al.
grâce au scanner 320-détecteurs : étude initiale

la qualité des reconstructions, notam- Conflits d’intérêt BEIR VII Phase 2. Washington, DC: Na-
ment sagittales, serait également un élé- tional Academies Press; 2006.
ment de comparaison intéressant à pren- Aucun. 14. Smith-Bindman R, Lipson J, Marcus R et
dre en compte pour l’évaluation de la al. Radiation dose associated with common
computed tomography examinations and
qualité d’image entre les deux groupes.
the associated lifetime attributable risk of
Troisièmement, notre étude a été réalisée
sur le même scanner. Il serait intéressant
Références cancer. Arch Intern Med 2009;169:2078-
86.
d’évaluer la dose délivrée pour un scanner 1. Valat JP. Epidémiologie des lombalgies. 15. Berrington de González A, Mahesh M,
lombaire sur des scanners hélicoïdaux Rev Rhum 1998;65:172-4. Kim KP et al. Projected cancer risks from
comportant un nombre différent de détec- 2. Gepner P. Lombalgies, aspects socio-éco- computed tomography scans performed
teurs (par exemple, un scanner 16-détecteurs) nomiques, épidémiologiques et médico- in the United States in 2007. Arch Intern
ou sur des machines de constructeurs légaux. Rev Rhum 1994;61:5-7. Med 2009;169:2071-7.
différents. Enfin, il existe quelques dif- 3. Boos N, Lander PH. Clinical efficacy of 16. Richards PJ, George J, Metelko M,
férences de paramétrage entre les deux imaging modalities in the diagnosis of Brown M. Spine Computed Tomogra-
low-back pain disorders. Eur Spine J phy Doses and Cancer Induction. Spine
modes d’acquisition. Ainsi, l’absence de
1996;5:2-22. 2010;35:430-3.
pitch en mode Wide volume (en l’absen-
4. Stieber J, Quirno M, Cunningham M, 17. Little MP, Wakeford R, Tawn EJ, Bouf-
ce de mouvement de la table lors de l’ac- Errico TJ, Bendo JA. The reliability of fler SD, Berrington de Gonzalez A. Risks
quisition des données, il n’y a pas de computed tomography and magnetic re- associated with low doses and low dose
pitch) peut faire discuter l’influence de sonance imaging grading of lumbar facet rates of ionizing radiation: why linearity
sa valeur en mode hélicoïdal. On peut arthropathy in total disc replacement pa- may be (almost) the best we can do. Ra-
également noter que l’on a utilisé des in- tients. Spine 2009;34:833-40. diology 2009;251:6-12.
dices de bruit légèrement différents en- 5. Lurie JD, Tosteson AN, Tosteson TD et 18. Spangfort EV. The lumbar disc hernia-
tre le groupe 1 (IB à 7,5) et le groupe 2 al. Reliability of readings of magnetic re- tion: a computer-aided analysis of 2,504
(IB à 6). Cette différence, liée au réglage sonance imaging features of lumbar spi- operations. Acta Orthop Scand Suppl
« empirique » de la qualité d’image, est nal stenosis. Spine 2008;33:1605-10. 1972;142:1-95.
toutefois faible et ne semble pas pouvoir 6. Carrino JA, Lurie JD, Tosteson AN et al. 19. Hourcade S, Treves R. Etude rétrospecti-
Lumbar spine: reliability of MR imaging ve concernant le scanner lombaire dans
expliquer l’écart significatif de dose en-
findings. Radiology 2009;250:161-70. les lombalgies et lombosciatiques dans le
tre les deux groupes. Par contre, notre
7. Arana E, Royuela A, Kovacs FM et al. département de le Haute-Vienne, à pro-
protocole ne comporte pas d’adaptation Lumbar spine: agreement in the interpre- pos de 132 cas. Rev Rhum 2002;69:589-96.
du kilo-Voltage en fonction du poids du tation of 1.5-T MR images by using the 20. Sodickson A, Baeyens PF, Andriole KP
patient. Ce paramètre était constant Nordic Modic Consensus Group Classifi- et al. Recurrent CT, cumulative radiation
dans notre étude afin de rendre les deux cation Form. Radiology 2010;254:809-17. exposure, and associated radiation-indu-
protocoles d’examens comparables (cela 8. Talbot A. Analyse des données relatives à ced cancer risks from CT of adults. Ra-
explique le faible effectif de notre étude la mise à jour des niveaux de référence diology 2009;251:175-84.
et l’inclusion de patients âgés de plus de diagnostiques en radiologie et en médeci- 21. Pasricha SS, Nandurkar D, Seneviratne
45 ans). Toutefois, en pratique courante, ne nucléaire. Rapport DRPH n°2008-02, SK et al. Image quality of coronary 320-
pour optimiser la réduction de la dose Institut de Radioprotection et de Sûreté MDCT in patients with atrial fibrillation:
Nucléaire. 2008, 126 p. initial experience. Am J Roentgenol
délivrée au patient, le kilo-Voltage de-
9. Tubiana, M, Aurengo A, Averbeck A. La 2009;193:1514-21.
vrait être adapté en fonction du morpho-
relation dose-effet et l’estimation des ef- 22. Torres FS, Crean AM, Nguyen ET et al.
type du patient, en accord avec le princi- fets cancérogènes des faibles doses de Abolition of respiratory-motion artifact
pe ALARA (53). rayonnements ionisants. Académie Na- in computed tomography coronary an-
tionale de Médicine, Institut de France– giography with ultrafast examinations: a
Académie des Sciences, Paris, 2005. comparison between 64-row and 320-row
10. Tubiana M, Feinendegen LE, Yang C, multidetector scanners. Can Assoc Radiol
Conclusion Kaminski JM. The linear no-threshold J 2010;61:5-12.
relationship is inconsistent with radiation 23. Brouwer PA, Bosman T, van Walder-
En conclusion, notre étude confirme que biologic and experimental data. Radiolo- veen MA, Krings T, Leroux AA, Willems
l’acquisition d’un scanner lombaire en gy 2009;251:13-22. PW. Dynamic 320-section CT angiogra-
mode Wide volume 320-détecteurs per- 11. Land CE. Low-dose extrapolation of ra- phy in cranial arteriovenous shunting le-
met de réduire significativement la dose diation health risks : some implications of sions. AJNR Am J Neuroradiol 2010;31:
délivrée au patient par rapport à l’acquisi- uncertainty for radiation protection at 767-70.
tion hélicoïdale 64-détecteurs. Le mode low doses. Health Phys 2009:97:407-15. 24. Page M, Nandurkar D, Crossett MP et al.
Wide volume 320-détecteurs a donc rem- 12. Brenner DJ, Doll R, Goodhead DT et al. Comparison of 4 cm Z-axis and 16 cm Z-
placé dans notre service le protocole héli- Cancer risks attributable to low doses of axis multidetector CT perfusion. Eur Ra-
ionizing radiation: assessing what we diol 2010;20:1508-14.
coïdal standard des scanners lombaires.
really know. Proc Natl Acad Sci USA 25. Kandel S, Kloeters C, Meyer H, Hein P,
D’autres études doivent être menées pour
2003;100:13761-6. Hilbig A, Rogalla P. Whole-organ perfu-
évaluer l’intérêt de ce nouveau mode 13. Board of Radiation Effects Research Di- sion of the pancreas using dynamic volu-
d’acquisition dans la réduction de la dose vision on Earth and Life Sciences Natio- me CT in patients with primary pancreas
délivrée au patient, notamment dans le nal Research Council of the National carcinoma: acquisition technique, post-
cadre des explorations thoraciques ou Academies. Health Risks From Exposure processing and initial results. Eur Radiol
abdomino-pelviennes. to Low Levels of Ionizing Radiation: 2009;19:2641-6.

J Radiol 2010;91

© 2016 Elsevier Masson SAS. Tous droits réservés. - Document téléchargé le 15/03/2016
 A Gervaise et al. Réduction de dose dans l’exploration du rachis lombaire 785
grâce au scanner 320-détecteurs : étude initiale

26. Kroft LJ, Roelofs JJ, Geleijns J. Scan time 35. Kalra KM, Maher MM, Toth TL, Ka- 44. Calzado A, Rodríguez R, Muñoz A.
and patient dose for thoracic imaging in math RS, Halpern EF, Saini S. Compari- Quality criteria implementation for brain
neonates and small children using axial son of Z-axis automatic tube current mo- and lumbar spine CT examinations. Br J
volumetric 320-detector row CT compa- dulation technique with fixed tube Radiol 2000;73:384-95.
red to helical 64-, 32-, and 16- detector current CT scanning of abdomen and 45. Ngaile JE, Msaki P, Kazema R. Towards
row CT acquisitions. Pediatr Radiol pelvis. Radiology 2004;232:347-53. establishment of the national reference
2009;40:294-300. 36. Décision du 22 novembre 2007 fixant les dose levels from computed tomography
27. Hein PA, Romano VC, Lembcke A, May modalités du contrôle de qualité des sca- examinations in Tanzania. J Radiol Prot
J, Rogalla P. Initial experience with a nographes. Journal Officiel de la Répu- 2006;26:213-25.
chest pain protocol using 320-slice volu- blique Française, 7 décembre 2007.
46. Biswas D, Bible JE, Bohan M, Simpson AK,
me MDCT. Eur Radiol 2009;19:1148-55. 37. Commission européenne. European gui-
Whang PG, Grauer JN. Radiation expo-
28. International Electrotechnical Commis- delines on quality criteria for computed
sure from musculoskeletal computerized
sion (IEC). Medical electrical equipment tomography. Publication EUR 16262
tomographic scans. J Bone Joint Surg Am
– Part 2-44: Particular requirements for EN. Bruxelles, Belgique : Commission
2009;91:1882-9.
the basic safety and essential performance européenne, 2000.
of X-ray equipment for computed tomo- 38. Mori S, Endo M, Nishizawa K et al. En- 47. Fazel R, Krumholz HM, Wang Y et al.
graphy, International Standard. Genève, larged longitudinal dose profiles in cone- Exposure to low-dose ionizing radiation
Suisse : IEC 60601-2-44, 2009. beam CT and the need for modified dosi- from medical imaging procedures. N En-
29. Tsedakis A, Damilakis J, Perisinakis K, metry. Med Phys 2005;32:1061-9. gl J Med 2009;361:849-57.
Stratakis J, Gourtsoyiannis N. The effect 39. Geleijns J, Salvadó Artells M, de Bruin 48. Heggie JC. Patient doses in multi-slice
of z overscanning on patient effective do- PW, Matter R, Muramatsu Y, McNitt- CT and the importance of optimisation.
se from multidectector helical computed Gray MF. Computed tomography dose Australas Phys Eng Sci Med 2005;28:86-
tomography examinations. Med Phys assessment for a 160 mm wide, 320 detec- 96.
2005;32:1621-9. tor row, cone beam CT scanner. Phys 49. International Commission on Radiologi-
30. van der Molen AJ, Geleijns J. Overran- Med Biol 2009;54:3141-59. cal Protection. 2007 recommendations of
ging in multisection CT : quantification 40. Task Group on Control of Radiation Do- the International Commission on Radio-
and relative contribution to dose - compa- se in Computed Tomography. Managing logical Protection (ICRP Publication
rison of four 16-section CT scanners. Ra- patient dose in computed tomography: a 103). Ann ICRP 2007;37:1-332.
diology 2007;242:208-16. report of the International Commission 50. Wintermark M, Maeder P, Verdun FR et
31. Cordoliani YS, Boyer B, Le Marec E, on Radiological Protection. Ann ICRP al. Using 80 kVp versus 120 kVp in perfu-
Jouan E, Hélie O, Beauvais H. Vade- 2000;30:7-45. sion CT measurement of regional cere-
mecum du scanner hélicoïdal : estimation 41. Huda W, Ogden KM, Khorasani MR. bral blood flow. AJNR Am J Neurora-
des doses, choix des paramètres. J Radiol Converting dose-length product to effec- diol 2000;21:1881-4.
2002;83:685-92. tive dose at CT. Radiology 2008;248:995-
51. Smith AB, Dillon WP, Lau BC et al. Ra-
32. Perisinakis K, Papadakis AE, Damilakis 1003.
diation dose reduction strategy for CT
J. The effect of x-ray beam quality and 42. Arrêté du 12 février 2004 relatif aux ni-
protocols: successful implementation in
geometry on radiation utilization effi- veaux de référence diagnostiques en ra-
neuroradiology section. Radiology 2008;
ciency in multidetector CT imaging. Med diologie et médecine nucléaire. Journal
Phys 2009;36:1258-66. Officiel de la République Française, 16 247:499-506.
33. Mori S, Nishizawa K, Kondo C, Ohno M, mars 2004. 52. Zarb F, Rainford L, McEntee MF. Image
Akahane K, Endo M. Effective doses in 43. Blum A, Noel A, Winninger D et al. quality assessment tools for optimization
subjects undergoing computed tomogra- Dose optimization and reduction in CT of CT images. Radiography 2010;16:91-
phy cardiac imaging with the 256-multis- of the musculoskeletal system including 168.
lice CT scanner. Eur J Radiol 2008; the spine. In: Tack D, Genevois PA, Ra- 53. McCollough C, Primak AN, Braun N,
65:442-8. diation Dose from Pediatric and Adult Kofler J, Yu L, Christner J. Strategies for
34. Blum A. Scanographie volumique multi- Multidetector Computed Tomography. reducing radiation dose in CT. Radiol
coupe. Paris : Masson, 2002. pp 39-62. Springer-Verlag eds 2007. pp 185-194. Clin North Am 2009;47:27-40.

J Radiol 2010;91

© 2016 Elsevier Masson SAS. Tous droits réservés. - Document téléchargé le 15/03/2016
98
 Chapitre 3

Article 2 : Amélioration de la qualité d’image scanographique en utilisant les reconstructions

itératives Adaptive Iterative Dose Reduction avec une acquisition wide-volume sur un scanner
320-détecteurs.

Gervaise A, Osemont B, Lecocq S, Micard E, Noel A, Felblinger J, Blum A. CT image quality improvement
using adaptive iterative dose reduction with wide-volume acquisition on 320-detector CT. Eur Radiol 2012; 22:
295-301.

99
100
Eur Radiol
DOI 10.1007/s00330-011-2271-7

COMPUTED TOMOGRAPHY

CT image quality improvement using adaptive iterative dose

reduction with wide-volume acquisition on 320-detector CT
Alban Gervaise & Benoît Osemont & Sophie Lecocq &
Alain Noel & Emilien Micard & Jacques Felblinger &
Alain Blum

Received: 14 July 2011 / Revised: 25 August 2011 / Accepted: 31 August 2011

# European Society of Radiology 2011

Abstract study. Again, FBP and AIDR images were compared.

Objectives To evaluate the impact of Adaptive Iterative Image noise and SNR were analysed using a Wilcoxon
Dose Reduction (AIDR) on image quality and radiation signed-rank test.
dose in phantom and patient studies. Results In the phantom, spatial resolution assessment
Methods A phantom was examined in volumetric mode on showed no significant difference between FBP and AIDR
a 320-detector CT at different tube currents from 25 to reconstructions. Image noise was lower with AIDR than
550 mAs. CT images were reconstructed with AIDR and with FBP images with a mean reduction of 40%. CNR and
with Filtered Back Projection (FBP) reconstruction algo- SNR were also improved with AIDR. In patients, quanti-
rithm. Image noise, Contrast-to-Noise Ratio (CNR), Signal- tative and subjective evaluation showed that image noise
to-Noise Ratio (SNR) and spatial resolution were compared was significantly lower with AIDR than with FBP. SNR
between FBP and AIDR images. AIDR was then tested on was also greater with AIDR than with FBP.
15 CT examinations of the lumbar spine in a prospective Conclusion Compared to traditional FBP reconstruction
techniques, AIDR significantly improves image quality
and has the potential to decrease radiation dose.
A. Gervaise (*) Key Points
Service d’Imagerie Médicale, & This study showed that Adaptive Iterative Dose Reduction
Hôpital d’Instruction des Armées Legouest, (AIDR) reduces image noise.
27 Avenue de Plantières,
57070 Metz, France & In a phantom image noise was reduced without altering
e-mail: [Link]@[Link] spatial resolution.
& In patients AIDR reduced the image noise in lumbar
A. Gervaise : B. Osemont : S. Lecocq : A. Blum spine CT.
Service d’Imagerie Guilloz, CHU NANCY,
29 avenue de Lattre de Tassigny, & AIDR can potentially reduce the dose for lumbar spine
54035 Nancy, France CT by 52%.

A. Noel Keywords Computed tomography . Multidetector CT .

CRAN UMR CNRS 7039, Centre Alexis Vautrin,
avenue de Bourgogne, Dose reduction . Image quality . Iterative reconstruction .
54511 Vandoeuvre-lès-Nancy, France Lumbar spine

E. Micard : J. Felblinger
CIT 801, INSERM, CHU Nancy,
allée du Morvan, Introduction
54511 Vandoeuvre-lès-Nancy, France
In the last decade, the number of CT examinations
E. Micard : J. Felblinger
performed has continually increased, reaching more than
IADI, U947, INSERM, CHU Nancy,
allée du Morvan, 68.7 million per year in the United States in 2007 [1, 2].
54511 Vandoeuvre-lès-Nancy, France While this increase is associated with significant improve-
 Eur Radiol

ments in diagnostic performance, it has also caused an contrast resolution (spatial resolution). The phantom was
increase in individual and collective radiation [3] and the positioned at the isocentre of the gantry.
potential associated risk of radiation-induced cancer [4, 5].
Thus, reducing the dose of radiation has become a major Acquisition protocol
concern. Despite the development of several technological
innovations, such as automatic tube current modulation [6– The protocol consisted in the acquisition of a volume
8], the development of dynamically adjustable z-axis X-ray covering the entire phantom (16 cm in the z-axis), with 13
beam collimation and the use of volumetric acquisition different tube current values (25, 50, 80 mA, followed by
mode to reduce over-ranging phenomenon [9, 10], dose 100 to 550 mA in increments of 50 mA). Because of the
reduction remains limited by the use of Filtered Back small diameter of the phantom (20 cm), the kilovolt peak
Projection (FBP) reconstructions. Indeed, these reconstruc- (kVp) was set at 100 kVp. Other acquisition parameters for
tions create a significant increase of image noise in the case the 13 acquisitions were: 1 s of rotation time, beam width
of excessive reduction of the dose [11]. Thanks to an of 320 detectors with 0.5 mm slice collimation, 512×512
important reduction in image noise, the recent development matrix, and 240 mm Field Of View (FOV).
of iterative reconstruction enables health practitioners to
solve this problem, thus allowing a significant reduction in CT image reconstructions
dose compared with FBP reconstructions [12].
In our institution, these iterative reconstructions have The 13 image sets were reconstructed using traditional 3D
been available since summer 2010 under Adaptive Iterative FBP reconstruction and AIDR iterative reconstruction with
Dose Reduction (AIDR) and can be applied to volumetric the same parameters (transverse 0.5 mm slice thickness
and wide-volume modes. AIDR constitutes a new recon- with 0.5 reconstruction interval, soft-tissue kernel FC 07).
struction algorithm based on statistical iterative reconstruc- Then, the 26 reconstructed volumes were sent and archived
tion techniques [13] and can be adapted for large cone- in the Picture Archiving and Communication System
beam CT examinations and for the particular three- (PACS) of our institution (Impax V5, ES; AGFA Technical
dimensional (3D) FBP used to reconstruct volumetric and Imaging Systems, Ridgefield, NJ, USA).
wide-volume acquisitions. The AIDR algorithm reduces
image noise through iteration loops in the reconstruction Evaluation of image quality
domain. Noise reduction is based on the comparison of
images reconstructed with a pre-established noise model. The image quality was evaluated from the measurement of
The final iterative image and the original image are image noise, SNR, CNR and spatial resolution. To assess
weighed and combined to create the AIDR image. The reproducible placement of a Region Of Interest (ROI) for
summation of the two images ensures that the spatial all acquisitions, a software based on Matlab™ (The Math-
resolution is preserved whilst the overall image noise is Works Inc.; Natick, Massachusetts, USA) was developed.
reduced. To our knowledge, there is no study on the Two ROIs (ROI #1 and #2) of 60 mm2 were placed in a
possible use of this algorithm on non-helical acquisitions. synchronised manner on the 26 volumes inside the phantom
The purpose of our study was to evaluate the impact of by a computer engineer (E. M.). ROI #1 was positioned in a
AIDR reconstructions on image quality and radiation dose 1% low-contrast target of 15 mm in diameter, and ROI #2
in phantom and patient studies. was located in the background area, adjacent to the
measured target (Fig. 1). Noise corresponded to the
measurement of the standard deviation of the measured
Materials and methods Hounsfield Units (HU) of ROI #2. The ratio between the
difference of mean attenuation values of these two ROIs
Phantom study (ROI #1 and ROI #2) and the standard deviation of ROI #2
corresponded to the CNR [14], while the ratio between the
Phantom mean attenuation value of ROI #2 and the standard
deviation of ROI #2 corresponded to the SNR [15].
A phantom (Catphan 500; The Phantom Laboratory, Salem, The spatial resolution of the image was assessed in a
NY, USA) was examined in volumetric mode with a 320- qualitative manner by a visual side-by-side comparison of
detector volume CT system (Aquilion ONE, Toshiba, the patterns of pairs of lines with the high-contrast
Japan). Using a low-contrast module CTP515 and a high- resolution module, between FBP and AIDR sets. Two
resolution module CTP528, we respectively evaluated low- radiologists (B.O. and S.L.) with six and eight respective
contrast resolution [image noise, Signal-to-Noise Ratio years of experience in reading CT examinations, performed
(SNR) and Contrast-to-Noise Ratio (CNR)] and high- the reading on sections with identical FOV and visualiza-
Eur Radiol

3D, Toshiba) with noise index set at 7.5 for a 5-mm slice
with soft filter and with minimum/maximum mA set at 100/
500, rotation time: 1 s, beam width: 320×0.5 mm).

Image reconstruction

The images were reconstructed using traditional 3D FBP

reconstructions and AIDR iterative reconstructions, with
the same reconstruction parameters (transverse slices of
0.5 mm every 0.5 mm, reconstruction filter FC07, identical
FOV between the two series). Once the reconstructions
completed, both sets of FBP and AIDR images were sent
and archived in our PACS.

Image quality evaluation

Fig. 1 Transverse CT slice of the phantom’s low-contrast module.
ROIs #1 and #2 were placed in the phantom to evaluate image noise, The image noise was evaluated in a quantitative manner by
CNR and SNR. ROI #1 was positioned in a 1% low-contrast target of measuring the standard deviation of a ROI rounded by
15 mm in diameter, and ROI #2 was placed in the background area, 100 mm2 and placed in a standardised manner by the same
adjacent to the measured target
radiologist (A.G.) in a homogeneous region of the left
psoas muscle at the level of the L5 pedicles, on an
tion window (Window/level setting: 400/50 HU). A transverse section of 0.5 mm in a soft-tissue window [6].
calculation of the modulation transfer function was also This step was performed with an Aquilion ONE CT
carried out at 50% and 5% between the FBP and AIDR Workstation (Display console version 4.62, Toshiba, Japan)
sets, respectively. This modulation transfer function was with synchronisation of the FBP and AIDR sets of each
computed as the angular mean of the two-dimensional patient (identical ROIs placed on a cut at the same level and
Fourier transform of the point-spread function, measured same place for each patient).
from CT images of a 0.28 mm tungsten wire centred in the The SNR was calculated for each ROI by dividing the
phantom. mean attenuation value by the standard deviation [15].
Two board-certified radiologists (B.O. and S.L.), with
Patient study six and eight respective years’ of experience in reading
musculoskeletal CT, then performed independent qualita-
Our institutional review board approved the study and tive image noise analysis. They were neither involved in
informed consent was obtained from all patients participating selecting patients nor in conducting CT examinations. They
in the study were blind to all patient identifiers. The visual perception of
noise, defined by the grainy appearance of the CT images,
Population studied was evaluated on 0.5 mm soft tissue transverse CT slices
viewed with a standard window setting (window/level: 400/
This prospective monocentric study included 15 consecu- 50 HU). The image noise was graded in three categories,
tive patients (6 men and 9 women, mean age of 53.6 years according to the degree of image noise (1 = minimal, 2 =
old) between December 2010 and February 2011 referred to moderate and 3 = significant). This grading was done on
us for lumbar CT without injection of any contrast agent, in PACS workstations (AGFA Impax ES review station,
search of a disco-radicular conflict. Exclusion criteria were AGFA Technical Imaging Systems) after a common training
patients under 18 years old, pregnant women, trauma and session. During the reading session, the FBP and AIDR
patients with prior lumbar surgery. datasets were presented randomly to the readers.

Acquisition protocol Statistical analysis

The lumbar spine examinations were performed on a 320-row All data were analysed with R for Windows (R Foundation
detector CT (Aquilion ONE, Toshiba, Japan), in wide-volume for Statistical Computing, Vienna, Austria). The image
mode, with an acquisition in two volumes covering the lumbar quality scores for qualitative image noise were recorded for
spine from T12 to S2, with identical acquisition parameters each radiologist, resulting in a total score for the assessment
(135 kVp, x,y,z-axis tube current modulation (SUREExposure of image noise for each reconstruction technique. A
 Eur Radiol

Wilcoxon signed-rank test was used to compare SNR, The scoring of qualitative image noise was also
CNR, quantitative and qualitative image noise evaluations significantly better for the AIDR set (1.33±0.7) compared
between both reconstruction modes. An interobserver to the FBP set (2.1±0.6) with a p<0.005. The calculation of
agreement for the two radiologists was estimated using the Kappa coefficient showed a good interobserver concor-
the Kappa test. A P value less than 0.05 was considered to dance (κ=0.74).
represent a statistically significant difference. The SNR was significantly improved with the AIDR
reconstructions compared to FBP reconstructions (2.36±0.6
vs. 3.50±0.9 respectively, with a P=0.003), which was
Results equivalent to a mean improvement of 47% (range 33–63%).

Phantom study
Discussion
There was a mean noise reduction of 40% (range 35–44%),
comparing AIDR and FBP reconstructions (Fig. 2). This Our phantom study confirms that at an equivalent dose and
noise reduction was accompanied by an improvement of the compared to the traditional 3D-FBP algorithm used for
SNR (mean improvement of 61%, range 56–67%) and the volumetric reconstruction, AIDR iterative reconstruction
CNR (mean improvement of 54%, range 33–67%). The generates a significant reduction of image noise. Interest-
percentage of noise reduction, SNR and CNR improvement ingly, the reduction level is equivalent, regardless of the
was constant, regardless of the initial level of image noise, initial quality of the original images. This means that AIDR
including low-dose acquisitions. reconstruction shows similar performances in terms of
However, there was no significant difference between noise reduction for images acquired at normal dose than
AIDR and FBP images, whether for the qualitative evaluation for noisy images acquired at low dose.
of spatial resolution (Fig. 3), or for the quantitative evaluation This noise reduction can be considered to either improve
based on MTF measurements (Fig. 4). the image quality by keeping the dose constant, or to reduce
the delivered dose while maintaining an equivalent image
Patient study (Table 1) quality. Given that there is a direct relationship between the
delivered dose (or milliamperage, in our study), and the
The quantitative measurement of image noise showed a image noise (when the radiation dose decreases by 1/c, the
significant reduction of noise on AIDR images (mean 15.6 image noise increases by the square root of c [16]), it is
±4 HU, range 11.1–26.6 HU) compared to FBP recon- possible to estimate the potential dose reduction based on
structions (mean 22.5±5 HU; range 16.6–36.2 HU), with a our noise reduction measurements. Indeed, in our phantom
p<0.001. This corresponds to a mean reduction of 31% of study, we calculated that a mean reduction of image noise
image noise with the AIDR (range 24–37%). by 40% was associated with a potential dose reduction of
about 64%. This estimate done with the phantom dose
reduction is comparable to phantom measurements provided
by the manufacturer.
Our phantom study also highlighted a significant
improvement of SNR and CNR values, thanks to AIDR
iterative reconstructions and compared to 3D-FBP recon-
structions. Indeed, in the absence of changes in attenuation
values during the process of reconstruction, noise reduction
is automatically accompanied by an improvement of these
parameters. At the same time, our study shows that the
significant noise reduction does not alter spatial resolution.
This point is critical because in regular clinical practice,
several other methods already enable the reduction of
image noise. However, these different techniques cause
either a degradation in spatial resolution (use of a “soft”
reconstruction filter, thickening of the slices [17]), or an
increase of the delivered dose (increase of kVp or mAs). On
Fig. 2 Graph showing image noise reduction with adaptive iterative
dose reduction (AIDR: red line) compared with the 3D-filtered back
the other hand, AIDR iterative reconstruction reduces noise
projection technique (FBP: blue line), at different tube currents in the without altering spatial resolution and without increasing
phantom study the dose. When considering other types of iterative
Eur Radiol

Fig. 3 Transverse CT image of

the high-contrast module with
3D-filtered back projection
(FBP) at 300 mAs (a) and
adaptive iterative dose reduction
(AIDR) at 100 mAs (b). Note
the same spatial resolution
assessed by high-contrast
objects and the nearly same
image noise

reconstruction, the data regarding spatial resolution alter- explained by the conceptual differences of these algorithms,
ation during reconstruction are variable. The phantom study which operate either in the image domain, in the raw-data
of Hara et al. [12] revealed a slight deterioration of the domain, or in both.
high-contrast image of an object using half-dose iterative Our patient-based study shows that AIDR allows a mean
ASIR (Adaptive Statistical Iterative Reconstruction) recon- reduction of image noise by 31% for lumbar spine CT
struction, compared to FBP reconstruction. The qualitative compared to 3D-FBP reconstruction. This noise reduction was
evaluation of spatial resolution on images of patients was significant both during quantitative (p<0.001) and qualitative
also the only parameter worse with ASIR, in comparison assessment by the two radiologists (p<0.005), despite the
with FBP reconstruction, without affecting the diagnostic small number of patients included. An extrapolation based
value of the image [12]. In their study, Prakash et al. on noise reduction allows an estimate of the potential dose
showed that ASIR-high definition reconstruction led to a reduction to about 52%. This major dose reduction potential
slight improvement in spatial resolution on phantom studies at equivalent image quality is even more valuable since
[18]. Finally, for Ghetti et al., IRIS (Image Reconstruction lumbar spine CT performed in a context of discoradicular
in Image Space) iterative reconstruction preserved spatial conflict detection, mainly concerns young patients [20] who
resolution during the reconstruction process in the course of are likely to undergo repeated examinations [3].
a phantom study [19]. These differences can probably be To our knowledge, this is the first study to evaluate the
performance of AIDR iterative reconstruction on patients
using 320 multidetector CT. Comparing our results with
other types of iterative reconstruction is difficult because

Table 1 Quantitative and qualitative image quality comparison in

lumbar spine CT with adaptive iterative dose reduction (AIDR) vs. 3D
filtered back projection (FBP)

FBP AIDR p valuea

Quantitative image quality evaluation

Noise (HU) 22.5±5 15.6±4 0.001
Signal to noise ratio 2.36±0.6 3.50±0.9 0.003
Qualitative image noise quality scoreb
Reader 1 2.07±0.5 1.33±0.5 0.005
Reader 2 2.13±0.3 1.33±0.5 0.001
Average 2.1±0.6 1.33±0.7 0.005
Fig. 4 Graph showing no significant difference in spatial resolution at
Except for p value, data are mean ± standard deviation
50% modulation transfer function (MTF) with adaptive iterative dose a
reduction (AIDR) compared with the 3D-filtered back projection Statistically significant for all parameters
b
(FBP) technique Grading scale for noise: 1 = minimal, 2 = moderate and 3 = important
 Eur Radiol

their implementation is different for each manufacturer. necessary to confirm our encouraging results. Moreover,
Indeed, with wide-volume acquisitions, enlarged cone angles the choice to study AIDR iterative reconstruction using
for volume scanning require reconstruction algorithms based lumbar spine CT examinations does not allow comparison
on a 3D FBP. Then, ASIR and IRIS iterative reconstructions with other types of iterative reconstruction, which have
respectively require the choice of a percentage of mixture been mainly studied using anatomic areas associated with
between FBP and ASIR images and the selection of a number higher-delivered doses or requiring follow-up studies, such
of iterations during the IRIS reconstruction process. The level as abdomen and chest CT [24]. This limitation is due to the
of dose reduction and the final image quality, depend on these fact that at the time of our study, AIDR iterative
parameters [15, 18, 21–23]. Choosing a too-large percentage reconstruction was only available in the volumetric and
of ASIR, or a too-high number of IRIS iterations, can cause wide-volume modes, which is the acquisition mode at
an alteration of the usual aspects of images with the which we performed our lumbar spine CT. The upcoming
emergence of the phenomenon of “oversmoothing”, due to installation of a new version, which will allow the
changes in the image noise spectra [11, 13]. This was reconstruction of acquired images in helical mode, will
particularly shown during the use of ASIR at 100%, which thus enable us to evaluate AIDR performance on thorax or
enables a dose reduction of more than 75%, but at the expense abdomen CT.
of the deleterious occurrence of this “oversmoothing” [21]. Second, in our patient study, we evaluated the improve-
The use of ASIR at 40% helps to maintain a correct aspect of ment of image quality and not the impact of AIDR iterative
the image, while allowing only a 50% dose reduction [12]. reconstruction on the diagnostic performance of the exam.
As far as AIDR iterative reconstructions are concerned, Moreover, we did not evaluate the influence of AIDR on
they automatically choose the number of iterations and bone structure. However, CT is preferred for the assessment
perform a mix between original and AIDR reconstructed of the bony structures of the spine [25]; hence numerous
images, allowing for a compromise between dose reduction lumbar examinations are still performed in our institution.
and maintenance of a typical image quality. This technical Nevertheless, the analysis of bone structures is less
specificity has the advantage of avoiding the choice of sensitive to noise variations than that of soft tissue and
parameters. However, it becomes impossible to study these alterations in bone structure often affect elderly patients, a
effects on dose reduction and image quality. In practice, population for which the risk of radiation is lower because
despite an important reduction in image noise, we have not of reduced life expectancy [26]. Moreover, during qualita-
experienced a significant change in their appearance tive assessment of image noise, a slight change in the
(Fig. 5). appearance of noise with AIDR reconstruction can bring a
Our study has several limitations. First, it is a prelimi- bias during the reading, with the recognition of AIDR
nary study involving a small number of patients, which compared to FBP sets.
indirectly evaluates the benefit of AIDR iterative recon- Additionally, we did not study the performances in terms
struction on dose reduction based on an extrapolation of of image quality improvement depending on the body
measurements of image noise reduction. A study comparing habitus of patients, and in particular the body mass index,
image quality between a scan at half-dose with the AIDR given the small number of patients included. Such a
and at normal dose with the FBP reconstruction is correlation would be of interest, though, to determine

Fig. 5 Transverse lumbar spine CT image reconstructed with change in image pattern (standard deviation values of the ROIs placed
traditional 3D-filtered back projection (FBP) (a) and adaptive iterative in left psoas are 21.1 HU with FBP and 14.5 HU with AIDR, which
dose reduction (AIDR) (b) in a 56-year-old woman. Note the noise corresponds to a noise reduction of 31%)
reduction with AIDR compared to FBP, without any significant
Eur Radiol

whether AIDR iterative reconstruction is as effective in 10. Gervaise A, Louis M, Batch T et al (2010) Dose reduction at CT
of the lumbar spine using a 320-detector scanner: initial results. J
patients who are overweight, where image quality is
Radiol 91:779–785
typically worse. Moreover, in our protocol, we did not 11. Xu J, Mahesh M, Tsui BM (2009) Is iterative reconstruction ready
adapt the kVp according to the body habitus of patients in for MDCT? J Am Coll Radiol 6:274–276
order to keep the acquisition parameters constant. 12. Hara AK, Paden RG, Silva AC, Kujak JL, Lawder HJ, Pavlicek W
(2009) Iterative reconstruction technique for reducing body
Finally, we did not analyse the time needed to reconstruct
radiation dose at CT: feasibility study. AJR Am J Roentgenol
images with AIDR. It does require more time than a standard 193:764–771
FBP reconstruction, but according to the vendor, AIDR 13. Thibault JB, Sauer KD, Bouman CA, Hsieh J (2007) A three-
reconstruction time per image is only 0.006 s longer than for dimensional statistical approach to improved image quality for
multislice helical CT. Med Phys 34:4526–4544
FBP reconstruction. In practice, when testing for a lumbar 14. Gupta AK, Nelson RC, Johnson GA, Paulson EK, Delong DM,
examination of 464 images, we effectively noticed a difference Yoshizumi TT (2003) Optimization of eight-element multi-detector
of approximately 2 s during the reconstruction process row helical CT technology for evaluation of the abdomen. Radiology
between FBP and AIDR (respectively 33 s vs. 35). 227:739–745
15. Pontana F, Duhamel A, Pagniez J et al (2011) Chest computed
In conclusion, our pilot study shows that compared to the
tomography using iterative reconstruction vs. filtered back
traditional FBP reconstruction technique, AIDR significantly projection (part 2): image quality of low-dose CT examinations
improves image quality in phantom and patient volumetric in 80 patients. Eur Radiol 21:636–643
and wide-volume CT. Further clinical evaluation is required to 16. Boedeker KL, McNitt-Gray MF (2007) Application of the noise
power spectrum in modern diagnostic MDCT: part II. Noise
confirm the radiation dose decreasing potential with AIDR.
power spectra and signal to noise. Phys Med Biol 52:4047–4061
17. von Falck C, Galanski M, Shin HO (2010) Informatics in
radiology: sliding-thin-slab averaging for improved depiction of
low-contrast lesions with radiation dose savings at thin-section
References CT. Radiographics 30:317–326
18. Prakash P, Kalra MK, Ackman JB et al (2010) Diffuse lung
disease: CT of the chest with adaptive statistical iterative
1. 2007 CT market summary report (2007) Des Plaines, IL: IMV reconstruction technique. Radiology 256:261–269
Medical Information Division 19. Ghetti C, Ortenzia O, Serreli G (2011) CT iterative reconstruction
2. Brenner DJ, Hall EJ (2007) Computed tomography an increasing in image space: a phantom study. Phys Med [Epub ahead of print]
source of radiation exposure. N Engl J Med 357:2277–2284 20. Spangfort EV (1972) The lumbar disc herniation: a computer-
3. Sodickson A, Baeyens PF, Andriole KP et al (2009) Recurrent CT, aided analysis of 2,504 operations. Acta Orthop Scand Suppl
cumulative radiation exposure and associated radiation-induced 142:1–95
cancer risks from CT of adults. Radiology 252:175–184 21. Silva A, Lawder H, Hara A, Kujak J, Pavlicek W (2010)
4. Smith-Bindman R, Lipson J, Marcus R et al (2009) Radiation Innovations in CT dose reduction strategy: application of the
dose associated with common computed tomography examina- adaptive statistical iterative reconstruction algorithm. AJR Am J
tions and the associated lifetime attributable risk of cancer. Arch Roentgenol 194:191–199
Intern Med 169:2078–2086 22. Pontana F, Pagniez J, Flohr T et al (2011) Chest computed
5. Berrington de González A, Mahesh M, Kim KP et al (2009) tomography using iterative reconstruction vs. filtered back
Projected cancer risks from computed tomography scans performed projection (part 1): evaluation of image noise reduction in 32
in the United States in 2007. Arch Intern Med 169:2071–2077 patients. Eur Radiol 21:627–635
6. Mulkens TH, Bellinck P, Baeyaert M et al (2005) Use of an 23. Mitsumori LM, Shuman WP, Busey JM, Kolokythas O, Koprowicz
automatic exposure control mechanism for dose optimization in KM (2011) Adaptive statistical iterative reconstruction versus filtered
multi-detector row CT examinations: clinical evaluation. Radiology back projection in the same patient: 64 channel liver CT image
237:213–223 quality and patient radiation dose. Eur Radiol [Epub ahead of print]
7. Kalender WA, Buchenau S, Deak P et al (2008) Technical 24. Fazel R, Krumholz HM, Wang Y et al (2009) Exposure to low-
approaches to the optimisation of CT. Phys Med 24:71–79 dose ionizing radiation from medical imaging procedures. N Engl
8. Alkadhi H, Leschka S (2011) Radiation dose of cardiac computed J Med 361:849–857
tomography—what has been achieved and what needs to be done. 25. Tins B (2010) Technical aspects of CT imaging of the spine.
Eur Radiol 21:505–509 Insights Imaging 1:349–359
9. Christner JA, Zavaletta VA, Eusemann CD, Walz-Flannigan AI, 26. Bohy P, de Maertelaer V, Roquigny A, Keyzer C, Tack D,
McCollough CH (2010) Dose reduction in helical CT: dynamically Genevois PA (2007) Multidetector CT in patients suspected of
adjustable z-axis X-ray beam collimation. AJR Am J Roentgenol having lumbar disk herniation: comparison of standard-dose and
194:W49–W55 simulated low-dose techniques. Radiology 244:231–524
108
 Chapitre 3

Article 3 : Réduction de la dose des scanners abdominopelviens grâce aux reconstructions

itératives AIDR 3D.

Gervaise A, Osemont B, Louis M, Lecocq S, Teixeira P, Blum A. Standard dose versus low-dose abdominal and
pelvic CT: comparison between filtered back projection versus adaptive iterative dose reduction 3D. Diagn
Interv Imaging 2014; 95: 47-53.

109
110
Diagnostic and Interventional Imaging (2014) 95, 47—53

ORIGINAL ARTICLE / Gastrointestinal imaging

Standard dose versus low-dose abdominal and pelvic

CT: Comparison between filtered back projection
versus adaptive iterative dose reduction 3D
A. Gervaise a,b,∗, B. Osemont a, M. Louis a, S. Lecocq a,
P. Teixeira a, A. Blum a

a
Service d’Imagerie Guilloz, Hôpital Central, CHU de Nancy, 29, avenue du
Maréchal-de-Lattre-de-Tassigny, 54035 Nancy cedex, France
b
Service d’Imagerie Médicale, Hôpital d’Instruction des Armées Legouest, 27, avenue de
Plantières, BP 90001, 57077 Metz cedex 3, France

KEYWORDS Abstract
Multidetector CT; Purpose: To compare the dose and image quality of a standard dose abdominal and pelvic CT
Dose reduction; with Filtered Back Projection (FBP) to low-dose CT with Adaptive Iterative Dose Reduction 3D
Image quality; (AIDR 3D).
Iterative Materials and methods: We retrospectively examined the images of 21 patients in the portal
reconstruction; phase of an abdominal and pelvic CT scan before and after implementation of AIDR 3D iterative
Abdominal and pelvic reconstruction. The acquisition length, dose and evaluations of the image quality were com-
CT pared between standard dose FBP images and low-dose images reconstructed with AIDR 3D and
FBP using the Wilcoxon test.
Results: The mean acquisition length was similar for both CT scans. There was a signifi-
cant dose reduction of 49.5% with low-dose CT compared to standard dose CT (mean DLP
of 451 [Link] versus 892 [Link], P < 0.001). There were no differences in image quality
scores between standard dose FBP and low-dose AIDR 3D images (4.6 ± 0.6 versus 4.4 ± 0.6
respectively, P = 0.147).
Conclusion: AIDR 3D iterative reconstruction enables a significant reduction in dose of 49.5%
to be achieved with abdominal CT scan compared to FBP, whilst maintaining equivalent image
quality.
© 2013 Éditions françaises de radiologie. Published by Elsevier Masson SAS. All rights reserved.

∗ Corresponding author. Service d’Imagerie Guilloz, Hôpital Central, CHU de Nancy, C29, avenue du Maréchal-de-Lattre-de-Tassigny, 54035
Nancy cedex, France.
E-mail address: [Link]@[Link] (A. Gervaise).

2211-5684/$ — see front matter © 2013 Éditions françaises de radiologie. Published by Elsevier Masson SAS. All rights reserved.
[Link]
48 A. Gervaise et al.

The number of CT scans performed has increased relent- Study population

lessly over the last decade, to more than 68.7 million
investigations in the United States in 2007 [1]. Although this This was a single-centre retrospective study which included
increase is associated with a large improvement in diagnos- 21 patients who had had contrast-enhanced abdominal and
tic performance, it is also a source of increased personal pelvic CT in the portal phase within our institution. Both
and population exposure to radiation [2,3] as well as being investigations were performed on the same machine (Aquil-
the source of a potential risk of radiation-induced can- ion ONE), the first using standard dose with standard FBP
cer from low doses of X-rays [4,5]. Reducing the radiation reconstructions (FBP standard dose group) and the sec-
dose delivered by CT scans has therefore become a major ond using low-dose with AIDR 3D iterative reconstructions
concern, particularly in abdominal imaging where the acqui- (AIDR 3D low-dose group). The images obtained with low-
sition protocol may include three or four acquisition phases dose were also reconstructed by FBP (FBP low-dose group).
[6]. Patients were identified and selected from our institution’s
The widespread use of several new technologies such PACS system (Picture Archiving and Communication System)
as automatic millamperage modulation [7—9] and active (Impax V5, ES; AGFA Technical Imaging Systems, Ridge-
collimation [10] has already reduced doses delivered in field, NJ, USA) by searching, during the 3 months after
abdominal and pelvic CT scans. This reduction, however, is the introduction of AIDR 3D iterative reconstructions (from
limited by the use of standard Filtered Back Projections (or September to November 2011), for all patients who had a
FBP) as these cause a significant increase in image noise if portal phase abdominal and pelvic CT scan and who had
the dose is reduced too much [11]. The recent emergence already had the same scan before the iterative reconstruct-
of iterative reconstructions solves this problem by greatly ions were introduced. The exclusion criteria were patients
reducing image noise and therefore allowing the dose to who were minors, pregnant women and those who had an
be significantly reduced in comparison with standard FBP interval of more than 18 months between the two scans.
reconstructions [12,13]. Patients’ weights were recorded at the time of the low-dose
A first version of the iterative reconstructions was mar- scan.
keted on the 320-detector CT scanner (Aquilion ONE,
Toshiba, Japan) in 2010 (Adaptive Iterative Dose Reduc- Acquisition protocol
tion, Toshiba, Japan). This model showed considerable
potential to improve image quality of the scanners and All of the abdominal and pelvic scans were obtained using
reduce their dose, although it had the disadvantage of only a 320-detector CT scan instrument (Aquilion ONE, Toshiba,
being available retrospectively and only for volume and not Japan) with acquisition covering the abdomen and pelvis
helicoidal-mode acquisitions [14,15]. A new, more sophis- from the bases of the lungs to the pubic symphysis. All of
ticated version of these iterative reconstructions has now the patients had at least one portal phase acquisition after
been marketed (Adaptive Iterative Dose Reduction 3D [AIDR intravenous administration of 150 mL of contrast medium
3D], Toshiba, Japan) and is available in our institution. This (OMNIPAQUE 350, GE Healthcare, Chalfont St. Giles, UK) via
can be used prospectively and for helicoidal acquisitions. peripheral infusion over a period of 70 seconds with flow
It has a more complicated algorithm than the first version rate of 4 mL/s. The acquisition parameters were identical
of the AIDR iterative reconstructions and contains iteration (Table 1) except for the automatic milliamperage modu-
loops in the image and raw data fields. To create the AIDR lation noise index in the three planes (SURE Exposure 3D,
3D image, the final iterative image and the original image Toshiba) which were set at 9 for a 5 mm section of the
are combined and weighted in order to ensure a reduc- ‘‘soft tissues’’ window for the standard dose scan and 10
tion in noise, at the same time preserving normal textures for the low-dose scan. With the introduction of the AIDR 3D,
and anatomical outlines. Initial tests on model systems on the automatic milliamperage modulation automatically and
our scanner showed that AIDR 3D iterative reconstructions prospectively reduces the exposure parameters.
could potentially reduce the radiation dose by nearly 50%.
AIDR 3D is now, therefore, the reconstruction mode used Image reconstruction
in routine clinical practice. We are not aware, however, of
any patient studies which have assessed the utility of this For the standard dose scans, the images were reconstructed
new reconstruction algorithm in order to reduce the radi- with FBP reconstructions (standard dose FBP group). Images
ation dose for abdominal and pelvic CT scans in clinical for the low-dose scans were reconstructed using AIDR 3D
practice. iterative reconstructions in ‘‘standard’’ mode (AIDR 3D low-
The aim of our study was to compare the dose and image dose group) and using FBP reconstructions (FBP low-dose
quality of an abdominal and pelvic CT scan using standard group). The three series of images were reconstructed in
dose FBP compared to an abdominal and pelvic CT scan with 2 mm transverse sections every 1.6 mm with an FC 07 recon-
low-dose and AIDR 3D iterative reconstructions and standard struction filter. All of the images were sent to and archived
FBP reconstructions in the same patient. in our PACS system.

Assessment of image quality

Materials and methods
Image noise was assessed quantitatively by measuring the
This single-centre study was approved by our regional ethics standard deviation of the regions of interest (ROI) located
committee. In view of the retrospective nature of the study, in the liver and aorta (Fig. 1). These regions were posi-
written consent from patients was not required. tioned in a standardised manner by the same radiologist
Standard dose versus low-dose abdominal and pelvic CT 49

Table 1 Scan acquisition and reconstruction settings.

Standard dose scan Low-dose scan

FBP AIDR 3D FBP

Acquisition mode Helicoidal Helicoidal Helicoidal
Detector collimation 64 × 0.5 mm 64 × 0.5 mm 64 × 0.5 mm
Kilovoltage 120 120 120
Sure Sure Sure
Milliamperage modulation Exposure 3D Exposure 3D Exposure 3D
Noise index 9 10 10
mA minimum/maximum 50/500 50/500 50/500
Section thickness/interval 2/1.6 mm 2/1.6 mm 2/1.6 mm
Reconstruction algorithm FBP AIDR 3D FBP
FBP: Filtered Back Projection; AIDR 3D: Adaptive Iterative Dose Reduction 3D; Sure Exposure3D: three dimensional automatic milliamperage
(mA) modulation (Toshiba, Japan).

(A.G.) on a 2 mm thick transverse section of the abdomen having a standard deviation of a 100 ± 20 mm2 region of
at the level of the portal vein bifurcation, from an Aquil- interest placed in the centre of the vessel in a homogenous
ion ONE post-treatment console (Display console, version region distant to the walls. For the liver, three circu-
4.74, Toshiba, Japan). For the aorta, noise was defined as lar 200 ± 50 mm2 regions of interest were positioned in

Figure 1. Abdominal CT images in a 26-year-old female patient being followed up for post-traumatic liver fracture. Two-millimetre
transverse CT sections of the abdomen at standard dose with FBP reconstructions (a) and low-dose CT with AIDR 3D iterative reconstructions
(b) and with FBP reconstructions (c). The position and size of the ROIs in the aorta (ROI #1) and in the liver (ROI #2—4) were maintained
between the three image series. Note the large reduction in image noise between the AIDR 3D low-dose (b) and FBP low-dose (c) groups.
Also note the similar image noise between the FBP standard dose (a) and AIDR 3D low-dose (b) groups, despite the large reduction in dose
with the second scan (529 [Link] vs. 267 [Link] respectively, i.e. a dose reduction of 50.5%).
50 A. Gervaise et al.

the left and right lobes of the liver, carefully avoiding Evaluation of transverse and anteroposterior
the intra-hepatic vessels, focal lesions and artefacts, as abdominal diameters
described by Marin et al. [16]. Liver noise was defined
as the mean of the standard deviations of the three liver Transverse and anteroposterior abdominal diameters (in cm)
ROIs. were measured for each patient on the standard dose and
In order to establish the independent relationship low-dose scans in order to ensure that there had been no
between image noise and dose for each of the three series significant change in patient body morphology between the
of images, a figure of merit (FOM) was calculated for the two investigations. These measurements were performed
aorta and for the liver using the equation described by Marin by the same radiologist (A.G.), on a PACS console, on the
et al. [16] where B2 is the square of the noise and ED is the same transverse sections passing through the portal vein
effective dose: FOM = 1/(B2 ·ED). bifurcation used to position the ROIs for the quantitative
Three senior radiologists (B.O., M.L. and A.B.) then per- measurement of image noise.
formed a qualitative assessment of image quality. These
radiologists have 6, 9 and 25 years of experience respec- Statistical analyses
tively in interpreting abdominal CT scans. They were not
involved in patient selection or in positioning the ROIs for Findings were analysed on R for Windows software (R Foun-
the quantitative image analysis. The images were read dation for Statistical Computing, Vienna, Austria). Mean
independently by the three radiologists on randomised, values were calculated for the image quality from the quan-
anonymised investigations which did not show the dates titative analysis by each radiologist to produce an overall
that the scans were performed. The assessment was image quality score for each group. The Wilcoxon signed
scored on a visual scale from 1 to 5 (1 = unacceptable rank test was used to compare acquisition lengths, doses
image quality, unable to interpret; 2 = poor image quality, delivered and abdominal diameters between the two types
interfering with interpretation; 3 = average image quality, of scan. The same test was used to compare the qualita-
interpretation possible; 4 = good image quality; 5 = excellent tive and quantitative assessments of image quality between
image quality) on PACS consoles after a joint reading the AIDR 3D low-dose, FBP low-dose and FBP standard dose
session. Image quality scores 1 and 2 were deemed groups. A P value of less than 0.05 was deemed to be a
to be unacceptable for interpretation in clinical prac- statistically significant difference.
tice.

Measurement of doses delivered Results

The doses delivered were provided directly from the inves- Twenty-one patients were included in the study (10 men and
tigation report which could be accessed in the PACS 11 women). The average age of the patients at the time of
system. They correspond to the CTDIvol (volume CT dose the low-dose scan was 43 ± 18 years (range: 21 to 86 years)
index) expressed in mGy and the Product Dose and Length and mean weight was 71 ± 8 kg (range: 45 to 88 kg). The
expressed in [Link]. The effective dose (ED) expressed in average interval between the two scans was 177 days (range:
milliSievert (mSv) was calculated using the tissue conversion 92 to 380 days). There were no significant differences in
coefficient (k) for the abdomen of 0.015 [17] by the equation acquisition lengths or abdominal diameters between the
ED = k × PDL [18]. low-dose and standard dose scans (Table 2).
Mean CTDIvol , DLP and effective doses of the low-
dose scans were significantly lower than with the standard
Evaluation of acquisition lengths dose scans (effective doses of 6.8 ± 2.5 mSv compared to
13.4 ± 4.3 mSv respectively, P < 0.001). The average reduc-
Acquisition length was expressed in centimetres and was tion in dose was 49.5% (Table 2).
measured as the difference in position between the first and For the quantitative assessment of image quality, we
last acquisition sections. found that mean image noise in the liver and aorta was

Table 2 Scan acquisition lengths, abdominal diameters and doses delivered.

Standard dose scan Low-dose scan P value
Acquisition length (cm) 45.6 ± 4 43.6 ± 2.6 0.07
Abdominal diameters (cm)
Anteroposterior 23.3 ± 2.7 23.6 ± 3.1 0.304
Transverse 31.7 ± 3.3 31.8 ± 3.2 0.667
CT radiation dose
CTDIvol (mGy) 19.5 ± 5.6 10.3 ± 3.6 < 0.001
DLP ([Link]) 892 ± 284 451 ± 170 < 0.001
Effective dose (mSv) 13.4 ± 4.3 6.8 ± 2.5 < 0.001
Apart from the P values, the results are expressed as mean ± standard deviation.
Standard dose versus low-dose abdominal and pelvic CT 51

Table 3 Quantitative and qualitative image quality assessment.

Standard dose Low-dose scan P value
scan
FBP AIDR 3D FBP AIDR 3D vs. AIDR 3D vs. FBP low-dose
FBP low-dose FBP standard vs. FBP
dose standard dose
Noise (UH)
Liver 15.86 ± 2.7 16.47 ± 1.4 27.21 ± 4.2 < 0.001 0.257 < 0.001
Aorta 17.62 ± 3.3 17.70 ± 2.5 31.84 ± 6.8 < 0.001 0.733 < 0.001
FOM (× 10−4 )
Liver 3.47 ± 1.0 6.27 ± 2.0 2.19 ± 1.0 < 0.001 < 0.001 < 0.001
Aorta 2.89 ± 1.0 5.71 ± 2.4 2.0 ± 1.0 < 0.001 < 0.001 0.003
Image quality score 4.6 ± 0.6 4.4 ± 0.6 3.3 ± 0.6 < 0.001 0.147 < 0.001
FBP: Filtered Back Projection; AIDR 3D: Adaptive Iterative Dose Reduction 3D. Apart from the P values, the results are expressed as
mean ± standard deviation.

significantly lower in the AIDR 3D low-dose group than in the diagnostic reference level defined in the 2012 legislation
the FBP low-dose group with noise reductions of 39 and 44% (800 [Link]) [19].
respectively. However, there was no significant difference The results of our study are similar to those of an initial
between mean liver or aortic image noise between the FBP study on the effectiveness of AIDR iterative reconstructions
standard dose and AIDR 3D low-dose groups (Table 3). The (Toshiba’s first version of iterative reconstructions) on lum-
mean of the FOMs in the liver and aorta were significantly bar spine CT scans which showed the potential to reduce the
higher for the AIDR 3D low-dose images compared to dose by 52% [14]. This dose reduction, however, was only
the FBP standard dose images (6.27 ± 2.0 compared to based on indirect calculation by extrapolating the reduction
3.47 ± 1.0, P < 0.001 for the liver and 5.71 ± 2.4 compared in noise from FBP to AIDR images on the same acquisition.
to 2.89 ± 1.0, P < 0.001 for the aorta, respectively) and FBP Our results are also similar to other types of itera-
low-dose (Table 3). tive reconstructions which have already been marketed and
The qualitative assessment of the image quality showed which are currently available in clinical practice [20—23].
this to be significantly higher in the FBP low-dose and Sagara et al. [20] and Prakash et al. [21] showed that it
AIDR 3D low-dose groups (3.3 ± 0.6 compared to 4.4 ± 0.6 was possible to reduce abdominal CT scan doses by 33% and
respectively, P < 0.001). There was no statistically signifi- 25% respectively, using ASIR, whilst improving image qual-
cant difference in mean image quality score between the ity in comparison with FBP reconstructed scans in patient
FBP standard dose and AIDR 3D low-dose groups (4.6 ± 0.6 studies using Adaptive Statistical Iterative Reconstruction
compared to 4.4 ± 0.6 respectively, P = 0.147) (Table 3). (ASIR). Mitsumori et al. [22] showed that the abdominal
scan dose could be reduced by 41% with ASIR in com-
parison with FBP reconstructions. In addition, as in our
Discussion own study, May et al. [23] demonstrated a 50% reduction
in abdominal scan dose using Iterative Reconstruction in
Our study confirms that the use of AIDR 3D iterative Image Space (IRIS) iterative reconstructions compared to
reconstructions greatly reduces image noise as compared standard FBP reconstructions, with equivalent image qual-
to standard FBP reconstructions. Our comparison between ity.
the FBP low-dose and AIDR 3D low-dose series images shows It is difficult, however, to compare our results with other
a significant improvement in subjective image quality and types of iterative reconstructions as their implantation is
in the quantitative assessment of image noise. Two of the different for each manufacturer. The ASIR and IRIS iterative
patients in the FBP low-dose group scored 2 out of 5 reconstructions, for example, respectively require a per-
for quality, thus the quality of the image interfered with centage mixing of FBP and ASIR images and a number of
interpretation, no patients in the AIDR 3D low-dose group iterations to be selected during the IRIS reconstruction pro-
scored 2 and only one of the 21 patients scored 3 in this cess. The dose reduction and quality of the final image both
group. depend on these parameters [20—24]. If an ASIR percentage
As a result of this reduction in image noise, it has become that is too high is chosen, or if too many iterations are used
possible to reduce acquisition parameters and therefore the for IRIS, changes may occur in the usual appearance of the
dose. Our comparison between the AIDR 3D low-dose and images with an ‘‘over-smoothing’’ effect due to a change in
FBP standard dose groups confirms that it is possible to halve the image noise spectra [11,12].
the radiation dose delivered in abdominal CT scans by using AIDR 3D also allows us to choose from four predetermined
AIDR 3D iterative reconstructions. This reduction in dose has modes: ‘‘weak’’, ‘‘mild’’, ‘‘standard’’ and ‘‘strong’’. These
enabled us to reduce the average PDL in our abdominal scans different modes allow a greater or lesser number of itera-
from 892 [Link] to 451 [Link], or a mean dose beneath tions to be performed and the mixing percentage of AIDR
52 A. Gervaise et al.

Figure 2. Low-dose abdominal CT in a 22-year-old female patient being followed up for post-traumatic splenic fracture
(DLP = 383 [Link]). Two-millimetre transverse sections centred on the pelvis using FBP reconstruction (a) and AIDR 3D iterative recon-
struction (b). Note the large reduction in image noise from the AIDR 3D iterative reconstructions (b) but also the reduction in metallic
artefacts from the intra-uterine device.

3D and FBP to be changed in the iterative reconstruction was a retrospective study, we did not have a record of the
process. The ‘‘standard’’ setting is the one recommended patient’s weight at their first scan and so we were not able
by the manufacturer for abdominal imaging and is a to compare their weights between the two investigations to
compromise between dose reduction and maintaining usual ensure that this had not changed significantly. Menke [26],
image quality. In practice, we have not noticed any differ- however, has shown that measurements of anteroposterior
ence in image texture on the AIDR 3D images and it was and transverse abdominal diameters correlate with patient
difficult for the readers to distinguish the FBP standard dose body morphology and particularly with body mass index.
and AIRD 3D low-dose images (Fig. 1). The ‘‘strong’’ setting The fact that these abdominal diameters did not change
may cause a slight change in usual image texture, although between the two scans in our study argues against a change
this setting can further reduce the radiation dose delivered. in body morphology in our patients. Finally, the dose reduc-
Yamada et al. showed that by using the ‘‘strong’’ setting tion found in our study only applies to abdominal CT scans
with AIDR 3D iterative reconstructions, the dose could be performed on our scanner and using our protocol. Other
reduced by 64% whilst maintaining equivalent image qual- studies are needed to assess the dose reduction for other
ity compared to standard FBP reconstructions in a study on reconstruction parameters, particularly with the ‘‘strong’’
chest CT scans [25]. Further studies are therefore needed setting, and also for other types of CT scan investigations,
to establish whether it is possible to use the ‘‘strong’’ set- particularly chest and brain.
ting in abdominal imaging in order to further decrease the
dose without reducing the diagnostic performance of the
investigations. Conclusion
Another advantage of some types of iterative reconstruc-
tion algorithms is that they reduce beam intensification AIDR 3D iterative reconstructions can halve the radiation
artefacts and metallic artefacts. AIDR 3D iterative recons- dose from an abdominal CT scan compared to standard FBP
tructions can partially correct these artefacts by using a reconstructions whilst maintaining equivalent image quality.
reconstruction algorithm with a double loop in the raw Further studies are needed to confirm the utility of itera-
data fields and in the image field. This partly explains the tive reconstructions in other types of CT scan investigations,
improvement in subjective image quality between FBP low- particularly chest and brain.
dose and AIDR low-dose images (Fig. 2).
There are several limitations to our study. Firstly, it is
a retrospective study which included a small number of
Disclosure of interest
patients. A larger-scale prospective study is needed in order
to confirm these results. Secondly, we only assessed image The authors declare that they have no conflicts of interest
quality and not the diagnostic performance of the scans concerning this article.
in our study. This would have been a more appropriate
style of assessment but it is difficult to implement. Thirdly,
we did not study the effect of patient body morphology
on the effectiveness of iterative reconstructions because
References
of the small numbers of patients included and as body
[1] 2007 CT market summary report. Des Plaines, IL: IMV Medical
mass index was not available for all patients. It would, Information Division.
however, be useful to carry out such a study with AIDR [2] Brenner DJ, Hall E. Computed tomography: an increasing
3D as several publications have shown that iterative source of radiation exposure. N Engl J Med 2007;357:2277—84.
reconstructions produce variable and contradictory results [3] Sodickson A, Baeyens PF, Andriole KP, Prevedello LM, Nawfel
depending on patient morphology [20,22]. Similarly, as this RD, Hanson R, et al. Recurrent CT, cumulative radiation
Standard dose versus low-dose abdominal and pelvic CT 53

exposure and associated radiation-induced cancer risks from [16] Marin D, Nelson RC, Schindera ST, Richard S, Youngblood
CT of adults. Radiology 2009;252:175—84. RS, Yoshizumi TT, et al. Low-tube-voltage, high-tube-current
[4] Smith-Bindman R, Lipson J, Marcus R, Kim KP, Mahesh M, multidetector abdominal CT: improved image quality and
Gould R, et al. Radiation dose associated with common com- decreased radiation dose with adaptive statistical iterative
puted tomography examinations and the associated lifetime reconstruction algorithm–initial clinical experience. Radiology
attributable risk of cancer. Arch Intern Med 2009;169:2078—86. 2010;254:145—53.
[5] Berrington de González A, Mahesh M, Kim KP, Bhargavan M, [17] Deak PD, Smal Y, Kalender WA. Multisection CT protocols: sex-
Lewis R, Mettler F, et al. Projected cancer risks from computed and age-specific conversion factor used to determine effective
tomography scans performed in the United States in 2007. Arch dose from Dose-Length-Product. Radiology 2010;257:158—66.
Intern Med 2009;169:2071—7. [18] International Commission on Radiological Protection. 2007 rec-
[6] Guite KM, Hinshaw JL, Ranallo FN, Lindstrom MJ, Lee FT. Ion- ommendations of the International Commission on Radiological
izing radiation in abdominal CT: unindicated multiphase scans Protection (ICRP Publication 103). Ann ICRP 2007;37:1—332.
are an important source of medically unnecessary exposure. J [19] Arrêté du 24 octobre 2011 relatif aux niveaux de références
Am Coll Radiol 2011;8:756—61. diagnostiques en radiologie et en médecine nucléaire. Journal
[7] Cordoliani YS, Boyer B, Le Marec E, Jouan E, Hélie O, Beauvais Officiel de la République Française, 14 January 2012.
H. Vade-mecum du scanner hélicoïdal : estimation des doses, [20] Sagara Y, Hara AK, Pavlicek W, Silva AC, Paden RG, Wu
choix des paramètres. J Radiol 2002;83:685—92. Q. Abdominal CT comparison of low-dose CT with adaptive
[8] Rizzo S, Kalra M, Schmidt B, Dalal T, Suess C, Flohr T, et al. statistical iterative reconstruction and routine-dose CT with
Comparison of angular and combined automatic tube current filtered back projection in 53 patients. AJR Am J Roentgenol
modulation techniques with constant tube current CT of the 2010;195:713—9.
abdomen and pelvis. AJR Am J Roentgenol 2006;186:673—9. [21] Prakash P, Kalra MK, Kambadakone AK, Pien H, Hsieh J, Blake
[9] Lee S, Yoon SW, Yoo SM. Comparison of image quality and MA, et al. Reducing abdominal CT radiation dose with adaptive
radiation dose between combined automatic tube current mod- statistical iterative reconstruction technique. Invest Radiol
ulation and fixed tube current technique in CT of the abdomen 2010;45:202—10.
and pelvis. Acta Radiol 2011;52:1101—6. [22] Mitsumori LM, Shuman WP, Busey JM, Kolokythas O, Koprowicz
[10] Christner JA, Zavaletta VA, Eusemann CD, Walz-Flannigan AI, KM. Adaptive statistical iterative reconstruction versus filtered
McCollough CH. Dose reduction in helical CT: dynamically back projection in the same patient: 64 channel liver CT
adjustable z-axis X-ray beam collimation. AJR Am J Roentgenol image quality and patient radiation dose. Eur Radiol 2012;22:
2010;194:W49—55. 138—43.
[11] Xu J, Mahesh M, Tsui BM. Is iterative reconstruction ready for [23] May M, Wust W, Brand M, Stahl C, Allmendinger T. Dose
MDCT? J Am Coll Radiol 2009;6:274—6. reduction in abdominal computed tomography: intra individ-
[12] Thibault JB, Sauer KD, Bouman CA, Hsieh J. A three- ual comparison of image quality of standard dose and half-dose
dimensional statistical approach to improved image quality for iterative reconstructions with dual-source computed tomogra-
multislice helical CT. Med Phys 2007;34:4526—44. phy. Invest Radiol 2011;46:465—70.
[13] Hara AK, Paden RG, Silva AC, Kujak JL, Lawder HJ, Pavlicek [24] Silva A, Lawder H, Hara A, Kujak J, Pavlicek W. Innovations in
W. Iterative reconstruction technique for reducing body radi- CT dose reduction strategy: application of the Adaptive Statis-
ation dose at CT: feasibility study. AJR Am J Roentgenol tical Iterative Reconstruction algorithm. AJR Am J Roentgenol
2009;193:764—71. 2010;194:191—9.
[14] Gervaise A, Osemont B, Lecocq S, Noel A, Micard E, Felblinger [25] Yamada Y, Jinzaki M, Hosokawa T, Tanami Y, Sugiura H, Abe T,
J, et al. CT image quality improvement using adaptive iterative et al. Dose reduction in chest CT: Comparison of the adaptive
dose reduction with wide-volume acquisition on 320-detector iterative dose reduction 3D, adaptive iterative dose reduction,
CT. Eur Radiol 2012;22:295—301. and filtered back projection reconstruction techniques. Eur J
[15] Tatsugami F, Matsuki M, Nakai G, Inada Y, Kanazawa S, Takeda Radiol 2012, [Link]
Y, et al. The effect of adaptive iterative dose reduction on [26] Menke J. Comparison of different body size parameters for
image quality in 320-detector row CT coronary angiography. Br individual dose adaptation in body CT of adults. Radiology
J Radiol 2012;85:e378—82. 2005;236:565—71.
118
 Chapitre 3

Article 4 : Scanner basse dose avec la modulation automatique du milliampérage, les

reconstructions Adaptive Statistical Iterative Reduction et un faible kilovoltage pour le
diagnostic des coliques néphrétiques : impact de l’indice de masse corporelle.

Gervaise A, Naulet P, Beuret F, Henry C, Pernin M, Portron Y, Lapierre-Combes M. Low-dose CT with

automatic tube current modulation, adaptive statistical iterative reconstruction, and low tube voltage for the
diagnosis of renal colic: impact of body mass index. Am J Roentgenol 2014; 202: 553-60.

119
120
 Genitourinar y Imaging • Original Research

Gervaise et al.
Impact of BMI on Low-Dose CT for Renal Colic

Genitourinary Imaging
Original Research
Low-Dose CT With Automatic
Tube Current Modulation, Adaptive
Statistical Iterative Reconstruction,
Downloaded from [Link] by [Link] on 10/13/15 from IP address [Link]. Copyright ARRS. For personal use only; all rights reserved

and Low Tube Voltage for the

Diagnosis of Renal Colic: Impact of
Body Mass Index
Alban Gervaise1 OBJECTIVE. The objective of our study was to evaluate the impact of body mass index
Pierre Naulet 1 (BMI) on dose, diagnostic performance, and image quality of a low-dose CT examination for
Florence Beuret 1 renal colic.
Christelle Henry 2 MATERIALS AND METHODS. This retrospective study included all patients who un-
Matthieu Pernin1 derwent a low-dose CT examination for renal colic performed during the year 2012 with au-
tomatic tube current modulation, adaptive statistical iterative reconstruction, and a low tube
Yann Portron1
voltage (kV). Three readers independently reviewed all images for the presence of renal colic
Marie Lapierre-Combes1 and evaluated diagnostic confidence and image quality. The results and doses were compared
Gervaise A, Naulet P, Beuret F, et al. among patients grouped by body mass index (BMI) and between patients with a BMI < 25
and those with a BMI t 25.
RESULTS. Eighty-six patients were included in the study: 39 patients had a BMI < 25
and 47 had a BMI t 25. No statistically significant difference was found between the accura-
cy rates for the diagnosis of renal colic when the rates of the three independent readers were
averaged for both BMI groups (95.7% vs 96.4%, respectively; p = 0.83). Image quality and
diagnostic confidence scores were significantly better for patients with a BMI t 25 than for
patients with a BMI < 25 (mean image quality score, 3.7 vs 3.4, p < 0.001; mean diagnostic
confidence score, 2.8 vs 2.5, p < 0.001). The mean radiation dose for patients with a BMI < 25
was 2.4 mGy compared with 3.7 mGy for patients with a BMI t 25 (p < 0.001).
CONCLUSION. The diagnostic performance of our low-dose CT protocol for renal
colic was excellent for all patients, and image quality and diagnostic confidence were signifi-
cantly better for patients with a BMI t 25. However, our protocol also required exposure to a
greater dose of radiation for these overweight and obese patients.

ver the past years, CT has become Taking into account the potential risks of

Keywords: body mass index, CT, iterative reconstruction,

O a reference technique in medical
imaging for the diagnosis of renal
colic [1, 2]. It is a fast examina-
radiation-induced cancer due to low x-ray
doses [5, 6] and the ALARA (as low as rea-
sonably achievable) precautionary princi-
tion that does not require the IV administra- ple, dose reduction of CT examinations per-
MDCT, radiation dose, renal colic
tion of contrast material. It also has excellent formed for suspected renal colic is essential.
DOI:10.2214/AJR.13.11350 performance for the diagnosis of renal colic Thanks to high contrast between most uri-
and enables the diagnosis of other diseases nary stones and the surrounding soft tissues,
Received June 5, 2013; accepted after revision and abnormalities clinically mimicking renal several authors have proposed low-dose CT
July 15, 2013.
colic. The main limitation of CT is that it in- protocols for patients with suspected renal
1
Department of Medical Imaging, Military Hospital volves an absorption of radiation dose, which colic using either a higher pitch or a sub-
Legouest, 27 Avenue de Plantières, BP 90001, Metz is especially limiting in this setting because stantially reduced tube current (mA) [7–13].
57077, Cedex 3, France. Address correspondence to urinary stone disease mainly affects young Many studies have shown that excellent di-
A. Gervaise ([Link]@[Link]).
people and tends to relapse. Leusmann et al. agnostic performance can be achieved with
2
Biochemistry Laboratory, University Hospital, Nancy, [3] evaluated the relapse rate of renal colic to low-dose CT and that diagnostic perfor-
France. be 35% at 10 years after diagnosis, and Katz mance is equivalent to standard-dose CT.
et al. [4] reported that 4% of patients undergo- However, Hamm et al. [9] and Poletti et al.
AJR 2014; 202:553–560 ing CT for suspected renal colic had already [12] found that the diagnostic performance
0361–803X/14/2023–553
undergone at least three CT examinations for of low-dose CT was decreased in obese pa-
the same indication, yielding cumulative dos- tients with a BMI > 30. That decrease in diag-
© American Roentgen Ray Society es ranging from 20 to 154 mSv. nostic performance resulted from the use of a

AJR:202, March 2014 553

 Gervaise et al.

constant tube current for all patients, leading automatic tube current modulation, an iter- Agfa Technical Imaging Systems) all the abdomi-
to a significant decrease in image quality for ative reconstruction such as adaptive statis- nopelvic CT examinations performed in 2012 and
the studies of obese patients. Thus, some au- tical iterative reconstruction (ASIR), and a by choosing only the unenhanced low-dose CT ex-
thors have proposed that low-dose CT not be low tube voltage while guaranteeing an ex- aminations of patients with suspected renal colic.
used for studies of obese patients [9, 10, 12], cellent diagnostic performance of low-dose CT examinations performed with other imaging
whereas others have recommended the use CT compared with standard-dose CT. How- settings and of patients referred for other indica-
of an adapted tube current for patients with ever, that initial study did not evaluate the tions (e.g., hematuria, follow-up CT for urolithi-
Downloaded from [Link] by [Link] on 10/13/15 from IP address [Link]. Copyright ARRS. For personal use only; all rights reserved

a BMI > 30 [13]. impact of patient morphology on image qual- asis) were not included in our study. One patient
Since these studies, the introduction of au- ity, diagnostic performance, and dose. was also excluded because of a technical issue dur-
tomatic tube current modulation during ac- Our study aimed at assessing the impact ing the low-dose CT acquisition.
quisitions has enabled tube current and im- of patient morphology on the dose, image For all patients, the following parameters were
age quality to be adapted to the patient’s quality, and diagnostic performance of our systematically recorded before the examination:
body habitus. Mulkens et al. [14] showed unenhanced low-dose CT protocol, which is sex, age (in years), weight (in kilograms), and
that it was possible to get excellent diagnos- performed with a low tube voltage, automat- height (in meters). For each patient, body mass in-
tic performance for CT studies of all patients ic tube current modulation, and ASIR, in pa- dex (BMI) was calculated as weight in kilograms
with suspected renal colic—including over- tients with suspected renal colic. divided by height in meters squared (weight /
weight patients and obese patients—by per- height2 [kg/m2]). Patients were classified accord-
forming low-dose CT examinations using Materials and Methods ing to their BMI as follows [19]: BMI < 18.5, thin;
automatic tube current modulation. This single-center study was approved by our BMI t 18.5 and < 25, normal weight; BMI t 25
The reduction of tube current remains lim- local ethics committee. This study is retrospec- and < 30, overweight; and BMI t 30, obese. Then
ited by the use of the standard filtered back tive, so the patients’ written informed consent was patients were divided in two groups according to
projection (FBP) reconstruction because the not necessary. their BMI: patients with a BMI < 25 and those with
FBP technique significantly increases the a BMI t 25.
image noise when the dose reduction is too Studied Population
important [15]. Recently, the development This retrospective single-center study includ- CT Acquisition and Reconstruction Techniques
of iterative reconstructions allowed a signifi- ed all patients who were referred to our imaging All examinations were performed with the pa-
cant noise reduction in CT images compared department for evaluation of suspected renal col- tient in a supine position using a 64-MDCT unit
with standard FBP reconstructions [16–18]. ic and underwent unenhanced abdominopelvic CT (Optima 660, GE Healthcare). Every examination
While maintaining the same image quali- performed with our routine low-dose protocol be- started with an acquisition of two scout views, a
ty, iterative reconstructions reduce noise so tween January 1, 2012, and December 31, 2012. lateral view and an anteroposterior view, using 120
that tube current and tube voltage can be re- Patients were referred by our emergency depart- kV and 10 mA. The low-dose CT protocol con-
duced, thus reducing the dose. In their pilot ment or by a physician from an outside institution. sisted of an unenhanced helical craniocaudal ac-
study, Kulkarni et al. [18] confirmed that it Patient selection was performed retrospectively by quisition that was centered on the urinary tract and
was possible to perform an acquisition with retrieving from our PACS (Impax ES, version 6, ranged from the upper pole of the kidneys (spotted
on the anteroposterior scout view) to the symphy-
sis pubis. The tube voltage was set at 100 kV, but
100.0 98.8 100.0
100 97.5 96.5
98.6
96.1
it remained possible to perform an acquisition us-
95.7
92.5 93.0 93.3 ing 120 kV for patients who weighed more than 80
90.0
90 kg and using 80 kV for patients who weighed less
than 60 kg. Other acquisition settings were con-
80
stant for all patients: automatic tube current modu-
Mean Performance Value (%)

70 lation in the x-axis, y-axis, and z-axis (Smart mA,

GE Healthcare) with a noise index setting of 50 for
60
a slice thickness of 1.25 mm and a soft-tissue win-
50 dow; tube current minimum and maximum settings
of 10 and 300 mA, respectively; a rotation time of
40 0.7 second; and a pitch of 1.375.
The noise index was set at 50 in consensus
30
with all the radiologists of our department. After
20 a 2-month experience (from November through
December 2011) using different noise index lev-
10 els, we chose the level that seemed to lead to the
best trade-off between dose reduction and suffi-
0
Reader 1 Reader 2 Reader 3 All Readers cient image quality for the diagnosis of renal colic.
Sensitivity Specificity Accuracy Images were reconstructed with ASIR (ASiR,
GE Healthcare) with a percentage setting of 50,
Fig. 1—Bar chart depicts mean sensitivity, specificity, and diagnostic accuracy of our low-dose CT protocol for a soft kernel, a slice thickness of 1.25 mm, and a
diagnosis of renal colic for each reader and all three readers combined. slice interval of 1.25 mm. The ASIR percentage

554 AJR:202, March 2014

 Impact of BMI on Low-Dose CT for Renal Colic

was the proportion of ASIR imaged within a mix scale (1 = no diagnostic confidence, 2 = confidence product (DLP), in mGy × cm, of the unenhanced
of standard FBP images and ASIR images. This with reservations, or 3 = total diagnostic confidence) low-dose CT examination. The effective dose
percentage was set according to the recommenda- and graded subjective image quality for the diagno- (ED), in mSv, was calculated using a tissue conver-
tions of the manufacturer. sis of renal colic on a 5-point Likert scale (1 = unac- sion coefficient (k) for the abdomen of 0.015 [20]
When the radiologist had any doubt about the ceptable image quality, 2 = suboptimal image qual- according to the following formula [21]:
presence of a stone after the unenhanced low-dose ity, 3 = acceptable image quality, 4 = good image
ED = k × DLP.
acquisition, the radiologist could perform another quality, or 5 = excellent image quality).
Downloaded from [Link] by [Link] on 10/13/15 from IP address [Link]. Copyright ARRS. For personal use only; all rights reserved

unenhanced acquisition with a standard dose. The To compare the average doses of our low-dose
standard-dose study used automatic tube current Final Diagnosis CT protocol and the standard-dose CT protocol,
modulation, a noise index setting of 21.5, and 120 For each examination, the final diagnosis was we compared the average DLP of the additional
kV, and the other acquisition and reconstruction set- established by another study investigator who had contrast acquisition with the average DLP of the
tings were the same as those for the low-dose study. 7 years of experience in abdominopelvic CT read- unenhanced low-dose acquisition for the patients
Similarly, if no renal colic was visible on the ing at the time of the study. He had access to the who underwent an additional contrast-enhanced
first acquisition and another diagnosis was sus- patient database, which included the unenhanced examination at the standard dose.
pected, the radiologist could perform an addition- low-dose CT study and potential additional ac-
al abdominopelvic acquisition during the portal quisitions with or without contrast material per- Statistical Analysis
venous phase with a standard dose to complete formed at a standard dose, and to the medical Data were analyzed using statistics software
the examination. The standard-dose study used a database, which included clinical, biologic, and (R Foundation for Statistical Computing version
noise index setting of 21.5 and 120 kV, and the follow-up information such as follow-up CT, com- 2.15.3) for Microsoft Windows. The sensitivity,
other acquisition settings were the same as those plementary ultrasound or MRI, reports of medical specificity, and diagnostic accuracy of the unen-
for the low-dose study. appointments, and surgery reports. hanced low-dose protocol were calculated for each
All images from the examination and the re- reader on the basis of the presence or absence of re-
port including the dosimetric data were sent di- Objective Evaluation of Image Noise nal colic compared with the final diagnosis given by
rectly to our PACS and were archived in the data- Image noise in unenhanced low-dose CT was the study investigator as a reference. Interobserver
base of our PACS. objectively evaluated by measuring the SD of the variability was calculated using Cohen’s kappa test.
attenuation (in Hounsfield units) in a region of in- Interobserver variability was considered excellent
Images Analysis terest (ROI) with an area of 100 mm 2. The ROIs for a kappa value of greater than 0.80. For the pa-
Unenhanced low-dose CT images were ana- were placed in the left psoas muscle, at the lev- tients who underwent an additional standard-dose
lyzed by two senior radiologists who had 8 and el of the fifth lumbar vertebra, in a standardized contrast acquisition, the average DLP of the low-
5 years of experience, respectively, in abdomino- way by the same investigator (investigator who es- dose CT study and the average DLP of the standard-
pelvic CT reading at the time of the study (readers tablished the final diagnosis) on a 1.25-mm-thick dose CT study were compared using the Wilcoxon
1 and 2) and by a third-year resident in radiology slice in soft-tissue window settings on a postpro- signed rank test. A correlation coefficient was calcu-
(reader 3). The readers were not involved in pa- cessing workstation (ADW 4.6). lated between the CTDI and BMI using a linear cor-
tient selection and had no information about clini- relation Pearson test. The CTDIs, averages of objec-
cal data or the final diagnosis. Evaluation of Radiation Dose tive image noise measurements, and average scores
The readers did not have access to the poten- Radiation dose data were directly provided in of image quality and diagnostic confidence (aver-
tial additional acquisitions performed at a stan- the examination report, which was accessible from age of the three readers’ scores) were compared for
dard dose with or without contrast material. The the PACS. The data consisted of the volume CT the two patient groups (BMI < 25 vs BMI t 25) us-
CT interpretations were performed independently dose index (CTDIvol), in mGy, and the dose-length ing a Mann-Whitney U test. The average diagnos-
by the three readers on randomized examinations
that had no patient-identifying information. Be-
fore interpreting the CT examinations, the read-
ers completed a common training session consist-
ing of 10 unenhanced low-dose abdominopelvic
CT studies performed in November or December
2011 that were not included in the study. Images
were visualized on a postprocessing workstation
(ADW, version 4.6, GE Healthcare). The readers
were allowed to use all the available visualiza-
tion tools. They could increase the slice thickness,
select display multiplanar reformations or maxi-
mum intensity projections, and use zoom. A B
During the reading session, the readers assessed
Fig. 2—41-year-old man (height, 1.78 m; weight, 78 kg; body mass index, 25.5) who presented with acute left-
the examinations for the presence or absence of re- sided lumbar pain. Unenhanced low-dose CT examination (100 kV; noise index, 50) with dose-length product of
nal colic. If the diagnosis was positive for renal col- 105 mGy × cm and effective dose of 1.6 mSv was performed.
ic, the readers recorded the size and localization of A, Axial image obtained with 1.25-mm slice thickness shows 2.5-mm urinary stone in pelvic ureter (arrowhead)
that led to false-negative finding of “small phlebolith” by all three readers.
the responsible stone. For all examinations, they B, Axial image obtained with 1.25-mm slice thickness shows small obstructive dilatation (arrow) without
graded diagnostic confidence on a 3-point Likert perirenal fat stranding.

AJR:202, March 2014 555

 Gervaise et al.

tic accuracy, number of false-positives, and number For 33 of the 86 patients (38.4%), a comple- in two patients, ileitis in one patient, diver-
false-negatives were compared using a Fisher test. mentary contrast acquisition was performed ticulitis in one patient, colitis in one patient,
A p value < 0.05 was considered to indicate a sig- during the portal venous phase using the stan- and stomach ulcer in one patient. Finally, for
nificant statistical difference. dard-dose protocol. There were 13 men and 20 the remaining 18 patients (21%) who did not
women with an average BMI of 24.1 ± 3.1 (SD) have renal colic, the causes of the painful
Results (range, 17.6–31.9). For two of the 86 (2.3%) symptoms could not be explained.
Patient and Acquisition Characteristics patients, an additional unenhanced acquisition
Downloaded from [Link] by [Link] on 10/13/15 from IP address [Link]. Copyright ARRS. For personal use only; all rights reserved

Eighty-six CT examinations of 86 patients was performed at the standard dose. There were Diagnostic Performance
were included in this study. There were 49 men two men with a BMI of 25.9 and 32. The sensitivity, specificity, and diagnostic
and 37 women. The average age of the study accuracy of the unenhanced low-dose proto-
group was 43.8 ± 14.7 (SD) years (range, 21– Final Diagnosis col for each of the three readers and the av-
82 years), and the average BMI was 25.3 ± 4.2 Of the 86 patients, 40 (46.5%) had a final erage of these values are displayed in Figure
(SD) (range, 15.6–42.9; median, 25.3). Based diagnosis of renal colic. The average size of 1. The interobserver variability for the diag-
on BMI, four patients (5%) were thin, 35 (41%) the stones was 4.4 ±2 (SD) mm (range, 2–10 nosis of renal colic between the readers was
had a normal body weight, 39 (45%) were over- mm). Eight stones were located in the lumbar excellent: Kappa was 0.88 between readers 1
weight, and eight (9%) were obese. Thirty-nine ureter, two at the cross of the lumbar ureter and 3 and between readers 2 and 3 and was
patients (45%) had a BMI < 25 and 47 (55%) with the iliac vessels, four in the pelvic ureter, 0.95 between readers 1 and 2.
had a BMI t 25. and 26 at the ureterovesical meatus. Among For the most experienced reader (reader 1),
Among the 86 low-dose CT examinations the 46 patients who did not have renal colic, the unenhanced low-dose protocol had a high
performed, the tube voltage was modified for a diagnosis was made in 28 (32.5%): discora- sensitivity and a high specificity, 97.5% and
six examinations: It was set at 80 kV for one CT dicular abnormalities in 10 patients, urinary 100%, respectively, and a diagnostic accura-
study (patient weight, 58 kg; BMI, 22.7) and at tract infections in seven patients, gynecolog- cy of 98.8%. Reader 1 made one interpretation
120 kV for five studies (two patients weighing ical infections in two patients, pyeloureter- mistake (a false-negative), reader 2 made three
81 kg with a BMI of 27.7 and 28; three patients al junction syndrome in one patient, renal in- mistakes (three false-negatives), and reader 3
weighing 84, 100, and 107 kg with a BMI of farction in one patient, renal colic cured at the made six mistakes (four false-negatives and two
29.5, 31.6, and 42.8, respectively). time of CT in one patient, acute appendicitis false-positives). In total, the three readers made
10 mistakes when interpreting seven CT exami-
nations (seven patients). One false-negative in-
terpretation of the same examination was made
by all three readers (Fig. 2); for another exami-
nation, the interpretations of both readers 2 and
3 were false-negative (Fig. 3). Of the five pa-
tients who had at least one false-negative inter-
pretation by one of the readers, three had a stone
enclosed in the ureterovesical meatus and two
had a stone in the pelvic ureter located just up-
stream from the ureterovesical meatus. These
five stones had a diameter of less than 3 mm.
A B In terms of diagnostic accuracy, false-neg-
atives, and false-positives, there was no sig-
nificant difference between the patients with a
BMI < 25 and those with a BMI t 25 (Table 1).
Of the 10 mistakes, five (three false-negatives
and two false-positives) were made interpret-
ing studies of patients with a BMI < 25 and five
(five false-negatives) of patients with a BMI
t 25. No mistake was made interpreting the
studies of the eight patients with a BMI t 30.

C D ,PDJH4XDOLW\DQG'LDJQRVWLF&RQÀGHQFH6FRUHV
The mean diagnostic confidence and image
Fig. 3—22-year-old woman (height, 1.72 m; weight, 62 kg; body mass index, 21) who presented with acute right- quality scores for all patients are recorded in
sided flank pain.
A and B, Axial unenhanced low-dose CT images obtained with 1.25-mm slice thickness (100 kV; noise index, 50;
Figure 4; these scores are also sorted accord-
dose-length product [DLP], 72 mGy × cm; effective dose [ED], 1.1 mSv) show 2-mm stone at junction of ureter ing to patient BMI and show that the higher the
with bladder (arrowhead, A) with minimal obstructive dilatation (arrow, B). Stone was misinterpreted as “small BMI, the higher the scores. The scores of di-
phlebolith” by two of three readers. agnostic confidence and of image quality were
C and D, Axial portal phase enhanced standard-dose CT images obtained with 1.25-mm slice thickness (120
kV; noise index, 21.5; DLP, 333 mGy × cm; ED, 5 mSv) show stone (arrowhead, C) depicted in A and allow better significantly better for patients with a BMI t 25
visualization of obstructive dilatation (arrow, D) in comparison with unenhanced low-dose CT image shown in B. than for those with a BMI < 25 (Table 1).

556 AJR:202, March 2014

 Impact of BMI on Low-Dose CT for Renal Colic

Objective Evaluation of Image Noise TABLE 1: Dose, Objective Image Noise Measurement, Image Quality and
The average of the objective image noise Diagnostic Confidence Scores, Diagnostic Accuracy, and False-
measurements for all patients was 33.6 ± 7.2 Positive and False-Negative Results for All Patients and Patient
(SD) HU. The distribution of noise values Groups Classified by Body Mass Indexa (BMI)
according to patient BMI is given in Table Patient or Study Characteristic All Patients BMI < 25 BMI ≥ 25 p
2. Concerning image noise measurements, No. (%) of patients 86 (100) 39 (45) 47 (55)
there was no significant difference between
Downloaded from [Link] by [Link] on 10/13/15 from IP address [Link]. Copyright ARRS. For personal use only; all rights reserved

CTDIvol (mGy) < 0.001

the patients with a BMI < 25 and those with
a BMI t 25 (Table 1). Mean ±SD 3.1 ±1.2 2.4 ±0.8 3.7 ±1.2
Range 1.4–8.1 1.4–4.3 2.0–8.1
Dose Evaluation Objective image noise (HU) 0.56
Average CTDIs, DLPs, and EDs for all
Mean ±SD 33.6 ±7.2 34.1 ±6.2 33.1 ±6.2
patients were 3.1 ±1.2 mGy (range, 1.4–8.1
mGy), 140.6 ±59.4 mGy × cm (range, 52.4– Image quality scoreb < 0.001
383.2 mGy × cm), and 2.1 ±0.9 mSv (range, Mean ±SD 3.6 ±0.7 3.4 ±0.6 3.7 ±0.6
0.8–5.7 mSv), respectively. The distribution Diagnostic confidence scorec < 0.001
of CTDIs according to patient BMI is given
Mean ±SD 2.6 ±0.6 2.5 ±0.6 2.8 ±0.4
in Table 2. There was an excellent correlation
between patient BMI and the CTDI of the low- Diagnostic accuracy (%) 96.1 95.7 96.4 0.83
dose CT examination: The Pearson linear cor- False-negative (no. of examinations)d 8 3 5 0.48
relation coefficient was 0.81 (Fig. 5). There False-positive (no. of examinations)d 2 2 0 0.14
was a significant difference between the radia-
Note—CTDIvol = Volume CT dose index.
tion dose of patients with a BMI < 25 and that aCalculated as weight in kilograms divided by height in meters squared.
of patients with a BMI t 25 (mean, 2.4 vs 3.7 bReaders graded subjective image quality for the diagnosis of renal colic on the following 5-point Likert scale:

mGy, respectively; p < 0.001) (Table 1). 1 = unacceptable image quality, 2 = suboptimal image quality, 3 = acceptable image quality, 4 = good image
quality, or 5 = excellent image quality.
Concerning the 33 examinations with an cReaders graded diagnostic confidence on the following 3-point Likert scale: 1 = no diagnostic confidence, 2 =
additional standard-dose contrast-enhanced confidence with reservations, or 3 = total diagnostic confidence.
d Average for all three readers.
CT study during the venous portal phase, the
average DLP of the contrast acquisitions was
583.2 ± 215.2 mGy × cm (range, 324.0–899.2 for the detection of renal colic was excellent. is no significant difference for patients with a
mGy × cm). There was a significant differ- The most experienced reader (reader 1) had BMI < 25 and those with a BMI t 25, with di-
ence between the dose of the unenhanced a diagnostic accuracy of 98.8%, and the av- agnostic accuracies of 95.7% and 96.4%, re-
low-dose CT examinations and that of the erage diagnostic accuracy for all three read- spectively. Our study shows that, thanks to
contrast-enhanced standard-dose CT exam- ers was 96.1%, which matches the data de- the use of automatic tube current modulation,
inations (140.6 vs 583.2 mGy × cm, respec- scribed in other studies [23]. Our results also it is possible to perform a low-dose CT pro-
tively; p < 0.001), with a 76% reduction in showed excellent interobserver concordance. tocol for suspected renal colic in overweight
dose for the low-dose CT protocol. As we expected, when comparing the dose patients and obese patients, which is consis-
results of the two groups of patients with a tent with the results reported by Mulkens et
Discussion BMI < 25 and BMI t 25, the dose for the CT al. [14]. However, the use of automatic tube
The results of our study confirm that the examinations of the overweight and obese pa- current modulation requires an increase in
use of our low-dose CT protocol for the diag- tients was significantly greater. This difference dose when imaging overweight patients and
nosis of renal colic leads to a 76% dose reduc- in dose by patient BMI is because of the use obese patients.
tion compared with our standard-dose con- of automatic tube current modulation, which It is interesting to note that, although the
trast acquisition (140.6 vs 583.2 mGy × cm, adapts the tube current to the patient’s body quantitative measurements of image noise re-
respectively). The average dose of the low- habitus. A higher tube current value is used mained stable regardless of patient BMI, the
dose CT protocol (2.1 mSv) was less than the for overweight patients and obese patients to scores of image quality and of diagnostic con-
average dose of an excretory urography ex- maintain image quality; this higher tube cur- fidence increased progressively with BMI.
amination (2.6 mSv [22]). This dose value of rent setting, therefore, leads to a significant in- Thus, the image quality and diagnostic confi-
2.1 mSv is also in accordance with the doses crease in the dose delivered to these patients. dence scores were significantly better for the
of low-dose CT examinations performed for In our study, there was an excellent correla- patients with a BMI t 25 than for the patients
the detection of renal colic described in oth- tion between the CTDIvol and BMI (Pearson with a BMI < 25. These results could seem
er studies, which ranged from 0.7 to 2.7 mSv linear correlation coefficient = 0.81), which is paradoxical, but we think that they can be ex-
[23]. Even the average dose for overweight pa- equivalent to those reported by Mulkens et al. plained by the fact that, at the same noise lev-
tients and obese patients (2.5 mSv) was less [14] (range of Pearson correlation coefficients, el, it is easier to diagnose renal colic when the
than the average dose of an excretory urogra- 0.85–0.88) who also used automatic tube cur- patient has a greater amount of intraabdomi-
phy examination (2.6 mSv [22]). rent modulation on a scanner made by anoth- nal and intrapelvic fat [24]. Indeed, it is easier
Despite this dose reduction, the diagnos- er manufacturer. Concerning the diagnostic to delimit the pathway of the ureters when fat
tic performance of our low-dose CT protocol performance of our low-dose protocol, there is present even if there is greater image noise.

AJR:202, March 2014 557

 Gervaise et al.

We also think that it is easier to see the indi-

rect signs of renal colic, such as perirenal in- 4.5
filtration or a rim sign, on images of obese pa-
4.0 3.9
tients. In our study, despite the fact that we
3.6 3.6
found the same number of reader interpreta- 3.5
3.5
tion errors for patients with a BMI < 25 as for 3.1
patients with a BMI t 25, none of the read-
Downloaded from [Link] by [Link] on 10/13/15 from IP address [Link]. Copyright ARRS. For personal use only; all rights reserved

3.0 2.8
er interpretation errors was made on exami- 2.7
2.6 2.6

Mean Score
nations of obese patients. Seven of the eight 2.5
2.2
false-negatives were because of the presence
of a pelvic phlebolith in patients with a BMI d 2.0
25.9. The two false-positive reader interpreta-
1.5
tion errors were also caused by difficulties dif-
ferentiating a phlebolith from a urinary stone
1.0
in two patients of normal weight (BMI of 22.5
and 23.0). These results are in accordance with 0.5
the study of Mulkens et al. [14] in which two
false-positives were two pelvic phleboliths in 0.0
All Patients Underweight Normal Weight Overweight Obese
two thin patients (BMI of 19.8 and 20.1).
In the first studies that used a reduced but Image quality Diagnostic confidence

constant tube current, diagnostic difficulties

mainly involved interpreting examinations of Fig. 4—Bar chart depicts mean scores of image quality and diagnostic confidence of unenhanced low-dose CT
obese patients [9, 12]. Our study shows that protocol for diagnosis of renal colic for all patients and for different categories of patients by body mass index
(BMI) (calculated as weight in kilograms divided by height in meters squared). For all examinations, readers
the use of automatic tube current modulation graded subjective image quality for diagnosis of renal colic on 5-point Likert scale (1 = unacceptable image
seems to lead to the opposite results: Diag- quality, 2 = suboptimal image quality, 3 = acceptable image quality, 4 = good image quality, 5 = excellent image
nostic difficulties mainly occurred when in- quality) and graded diagnostic confidence on 3-point Likert scale (1 = no diagnostic confidence, 2 = confidence
terpreting examinations of thin patients. We with reservations, 3 = total diagnostic confidence).
used automatic tube current modulation,
which requires a configuration of a noise in- quality that is good enough for the diagnosis Our study also confirms that it is possible
dex and maximal and minimal tube current of renal colic. In a pilot study in which they to perform an acquisition with 100 kV for
thresholds. For our low-dose CT protocol, analyzed the use of automatic tube current most patients including overweight patients
the maximal threshold was set to 300 mA modulation with ASIR and a low tube volt- and obese patients. Reducing the tube volt-
(i.e., 152 mAs). Given the better image qual- age, Kulkarni et al. [18] had indeed set the age from 120 to 100 kV leads to a dose re-
ity for patients with a BMI t 25, we think that maximal tube current–exposure time thresh- duction of almost 35%, but it also leads to
it would be possible to reduce this maximal old at 55 mAs with a noise index similar to an increase in the noise of approximately
threshold to reduce the dose for obese and ours. Other studies are necessary to evaluate 30% [25]. Thanks to an important reduc-
overweight patients while preserving image the optimal maximal tube current threshold. tion in image noise while keeping an equiv-

TABLE 2: Dose Data and Objective Image Noise Measurements for Unenhanced Low-Dose CT Examinations of
Patients Classified by Body Mass Indexa (BMI)
Patient or Study Characteristic BMI < 18.5 (Thin) 18.5 ≤ BMI < 25 (Normal Weight) 25 ≤ BMI < 30 (Overweight) BMI ≥ 30 (Obese)
No. (%) of patients 4 (5) 35 (41) 39 (45) 8 (9)
CTDIvol (mGy)
Mean ±SD 2.0 ±0.6 2.4 ±0.8 3.4 ±0.8 5.1 ±1.9
Range 1.4–2.7 1.5–4.3 2.1–5.1 3–8.1
Dose-length product (mGy × cm)
Mean ±SD 79.2 ±22.8 109.7 ±35.3 155.0 ±41.3 236.2 ±93.6
Range 52.4–106.6 58.6–209.3 92.5–263.1 124.7–383.2
Effective dose (mSv)
Mean ±SD 1.2 ±0.3 1.6 ±0.5 2.3 ±0.6 3.5 ±1.4
Range 0.8–1.6 0.9–3.1 1.4–3.9 1.9–5.7
Objective image noise (HU)
Mean ±SD 34.1 ±7.2 34.1 ±6.2 33.3 ±4.7 32.1 ±6.4
Note—CTDIvol = Volume CT dose index.
aCalculated as weight in kilograms divided by height in meters squared.

558 AJR:202, March 2014

 Impact of BMI on Low-Dose CT for Renal Colic

References
9
1. Smith RC, Rosenfield AT, Choe KA, et al. Acute
8 flank pain: comparison of non-contrast-enhanced
CT and intravenous urography. Radiology 1995;
7
194:789–794
6 2. Sung MK, Singh S, Kalra MK. Current status of
CTDI (mGy)

low dose multi-detector CT in the urinary tract.

Downloaded from [Link] by [Link] on 10/13/15 from IP address [Link]. Copyright ARRS. For personal use only; all rights reserved

5
World J Radiol 2011; 3:256–265
4 3. Leusmann DB, Niggemann H, Roth S, von Ahlen
H. Recurrence rates and severity of urinary cal-
3
culi. Scand J Urol Nephrol 1995; 29:279–283
2 4. Katz SI, Saluja S, Brink JA, Forman HP. Radia-
tion dose associated with unenhanced CT for sus-
1
pected renal colic: impact of repetitive studies.
0 AJR 2006; 186:1120–1124
0 5 10 15 20 25 30 35 40 45 50
5. Smith-Bindman R, Lipson J, Marcus R, et al. Ra-
BMI
diation dose associated with common computed
tomography examinations and the associated life-
Fig. 5—Graph shows excellent correlation between body mass index (BMI), calculated as weight in kilograms
divided by height in meters squared, and CT dose index (CTDIvol) of unenhanced low-dose CT: Pearson linear time attributable risk of cancer. Arch Intern Med
correlation coefficient was 0.81. 2009; 169:2078–2086
6. Berrington de González A, Mahesh M, Kim KP,
alent dose [17], the iterative reconstructions try to assess the performance of our low-dose et al. Projected cancer risks from computed to-
make acquisitions at 100 kV feasible without CT protocol for the detection of a diagnosis mography scans performed in the United States in
any important increase in image noise. Only because that was not the aim of our study. 2007. Arch Intern Med 2009; 169:2071–2077
five of the 13 patients weighing more than Moreover, when no abnormality was detect- 7. Diel J, Perlmutter S, Venkataramanan N, Mueller
80 kg in our study underwent low-dose CT ed on the low-dose CT study, most of our pa- R, Lane MJ, Katz DS. Unenhanced helical CT us-
with 120 kV. These cases show that at 100 kV tients underwent a complementary standard- ing increased pitch for suspected renal colic: an
the image quality and the diagnostic perfor- dose contrast-enhanced CT study. We think effective technique for radiation dose reduction? J
mance remain acceptable even for studies of that contrast administration in this setting is Comput Assist Tomogr 2000; 24:795–801
most obese and overweight patients. An even necessary because even an unenhanced stan- 8. Liu W, Esler SJ, Kenny BJ, Goh RH, Rainbow AJ,
greater tube voltage reduction can be con- dard-dose CT study with normal findings Stevenson GW. Low dose nonenhanced helical
templated: Kulkarni et al. [18] reported good does not allow pathologic entities such as re- CT of renal colic: assessment of ureteric stone de-
image quality using a low tube voltage set at nal infarcts to be eliminated from the differ- tection and measurement of effective dose equiva-
80 kV in patients who weighed less than 90 ential diagnosis; most pathologic entities that lent. Radiology 2000; 215:51–54
kg and at 100 kV in patients who weighed are diagnosed thanks to unenhanced CT re- 9. Hamm M, Knopfle E, Wartenberg S, Wawroschek F,
more than 90 kg. However, the possibility of quire a complementary contrast acquisition. Weckermann D, Harzmann R. Low dose unen-
using 80 kV for some categories of patients Finally, our results are valid for only an ac- hanced helical computed tomography for the evalua-
must be confirmed by other studies based on quisition performed on the CT unit used for tion of acute flank pain. J Urol 2002; 167:1687–1691
BMI and not on weight. our study with our specific automatic tube 10. Heneghan JP, McGuire KA, Leder RA, Delong
There are several limitations to our study current modulation software. Other studies DM, Yoshizumi T, Nelson RC. Helical CT for
that need to be mentioned. First, this study are necessary to generalize these results to nephrolithiasis and ureterolithiasis: comparison
was retrospective and included a relatively CT examinations performed on units made of conventional and reduced radiation dose tech-
low number of patients, especially obese pa- by other manufacturers and also to analyze niques. Radiology 2003; 229:575–580
tients. A prospective study of a larger num- the effect of some settings, especially the 11. Kim BS, Hwang IK, Choi YW, et al. Low-dose
ber of patients must be performed to confirm maximal tube current threshold suggested, and standard-dose unenhanced helical computed
our results. Second, there was no reference on dose and on diagnostic performance of tomography for the assessment of acute renal col-
imaging technique that could give a final di- low-dose CT. ic: prospective comparative study. Acta Radiol
agnosis for all of our patients, like could have In conclusion, our unenhanced low-dose 2005; 46:756–763
been done thanks to a standard-dose acqui- CT protocol performed with automatic tube 12. Poletti PA, Platon A, Rutschmann OT, Schmidlin
sition. This method does not seem ethical- current modulation, a low tube voltage set- FR, Iselin CE, Becker CD. Low-dose versus stan-
ly acceptable to us because it would have ting, and the ASIR iterative reconstruction dard-dose CT protocol in patients with clinically
required additional irradiation of a popula- for patients with suspected renal colic has suspected renal colic. AJR 2007; 188:927–933
tion that includes a large proportion of young an excellent diagnostic performance for all 13. Tack D, Sourtzis S, Delpierre I, de Maertelaer V,
or healthy patients. However, the reference patients. However, better image quality and Gevenois PA. Low-dose unenhanced multidetec-
method to make the final diagnosis that was better diagnostic confidence were achieved tor CT of patients with suspected renal colic. AJR
chosen, one that relies on all the imaging and for patients with a BMI t 25, despite a great- 2003; 180:305–311
patient-tracking data, has been used in sev- er radiation dose, compared with patients 14. Mulkens TH, Daineffe S, De Wijngaert R, et al.
eral studies [7, 13, 14, 26]. Third, we did not with a BMI < 25. Urinary stone disease: comparison of standard-

AJR:202, March 2014 559

 Gervaise et al.

dose and low-dose with 4D MDCT tube current for evaluation of urolithiasis: how low can we in intra-venous urography. Br J Radiol 2001;
modulation. AJR 2007; 188:553–562 go? Radiology 2012; 265:158–166 74:727–734
15. Xu J, Mahesh M, Tsui BM. Is iterative reconstruc- 19. Garrow JS, Webster J. Quetelet’s index (W/H2) as 23. Niemann T, Kollmann T, Bongartz G. Diagnostic
tion ready for MDCT? J Am Coll Radiol 2009; a measure of fatness. Int J Obes 1985; 9:147–153 performance of low-dose CT for detection of uro-
6:274–276 20. Deak PD, Smal Y, Kalender WA. Multisection CT lithiasis: a meta-analysis. AJR 2008; 191:396–401
16. Thibault JB, Sauer KD, Bouman CA, Hsieh J. A protocols: sex- and age-specific conversion factor 24. Katz DS, Venkataramanan N, Napel S, Sommer
three-dimensional statistical approach to im- used to determine effective dose from dose-length FG. Can low dose unenhanced multidetector CT
Downloaded from [Link] by [Link] on 10/13/15 from IP address [Link]. Copyright ARRS. For personal use only; all rights reserved

proved image quality for multislice helical CT. product. Radiology 2010; 257:158–166 be used for routine evaluation of suspected renal
Med Phys 2007; 34:4526–4544 21. International Commission on Radiological Pro- colic? AJR 2003; 180:313–315
17. Hara AK, Paden RG, Silva AC, Kujak JL, Lawder tection. The 2007 recommendations of the Inter- 25. McNitt-Gray MF. AAPM/RSNA physics tutorial
HJ, Pavlicek W. Iterative reconstruction technique national Commission on Radiological Protection: for residents: topics in CT—radiation dose in CT.
for reducing body radiation dose at CT: feasibility ICRP Publication 103. Ann ICRP 2007; 37:1–332 RadioGraphics 2002; 22:1541–1553
study. AJR 2009; 193:764–771 22. Yakoumakis E, Tsalafoutas IA, Nikolaou D, 26. Kluner C, Hein PA, Gralla O, et al. Does ultra-
18. Kulkarni NM, Uppot RN, Eisner BH, Sahani DV. Nazos I, Koulentianos E, Proukakis C. Differ- low-dose CT with a radiation dose equivalent to
Radiation dose reduction at multidetector CT ences in effective dose estimation from dose-area that of KUB suffice to detect renal and ureteral
with adaptive statistical iterative reconstruction product and entrance surface dose measurements calculi? J Comput Assist Tomogr 2006; 30:44–50

F O R YO U R I N F O R M AT I O N
See the AJR’s monthly Journal Club article in this issue on page 493.

560 AJR:202, March 2014

Discussion et conclusion du chapitre 3 :

Les résultats présentés dans ce chapitre montrent comment il est possible de réduire la
dose des scanners en optimisant certains paramètres d’acquisition et de reconstruction des
images.

Article 1 : Réduction de dose dans l’exploration du rachis lombaire grâce au scanner

320-détecteurs : étude initiale.

Notre étude montre que le mode d’acquisition peut avoir un impact non négligeable sur la
dose des scanners. Lors de l’acquisition de scanners lombaires, nous avons retrouvé une
réduction de la dose de 35 % en faveur du mode d’acquisition volumique par rapport au mode
d’acquisition hélicoïdal classique, à qualité d’image constante. Ce résultat est principalement
dû à la suppression de l’irradiation pré et post-hélice liée au phénomène d’overranging en
mode hélicoïdal. Par contre, compte-tenu de la longueur d’acquisition importante des scanners
lombaires (26 cm), il était nécessaire de réaliser une acquisition de deux volumes successifs.
Cette juxtaposition de deux volumes engendre une irradiation supplémentaire à la zone de
chevauchement des deux volumes et peut aussi être à l’origine d’un artéfact de décalage sur
les images du scanner à la zone de jonction des deux volumes. Ainsi, l’acquisition volumique
sera surtout préférée en cas d’acquisition pour des volumes avec une couverture d’acquisition
inférieure à 16 cm. Cette étude était une étude initiale. Nous avons essayé de confirmer ces
résultats par une nouvelle étude prospective de plus grande envergure mais cette dernière n’a
malheureusement pas aboutit. Les résultats intermédiaires ne retrouvaient pas un niveau de
réduction de la dose aussi important que dans notre étude initiale. Cette différence peut être
liée à plusieurs facteurs. Tout d’abord, dans notre étude initiale, l’indice de bruit de la
modulation automatique du mA était légèrement supérieur pour l’acquisition volumique par
rapport à l’acquisition hélicoïdale (7,5 versus 6). Cette différence peut expliquer une partie de
l’écart de dose entre les deux protocoles. De plus, lors de notre nouvelle étude, un
changement de version du logiciel de reconstruction des images a pu également influencer la
dose et la qualité des images des scanners. Enfin, l’évaluation de la qualité d’image de notre
étude initiale n’était peut-être pas assez pertinente et compte-tenu d’un faible nombre de
patients inclus, l’étude initiale n’a peut-être pas permis de montrer une différence significative
de qualité d’image alors que celle-ci existait peut-être. Même si l’étude complémentaire n’a

129
pas permis de retrouver des résultats aussi positifs que dans l’étude initiale, nos résultats ont
toutefois été confirmés par une étude portant sur la comparaison des modes d’acquisition
volumique et hélicoïdal avec le scanner 320-détecteurs dans une population pédiatrique. Kroft
LJ et al. ont effectivement montré une réduction de la dose au profit du mode volumique [48].
Cette réduction était d’autant plus importante que la longueur d’acquisition était petite. Ainsi,
pour une longueur d’acquisition de 16 cm, il retrouvait une réduction de 18 % de la dose en
faveur du mode volumique par rapport au mode hélicoïdal et pour une acquisition de 80 mm,
une réduction de 40 % de la dose, à qualité d’image équivalente. Un autre avantage de
l’acquisition volumique était la rapidité d’acquisition du volume (0,35 s pour l’acquisition
d’un volume de 16 cm), ce qui permettait de réduire de manière importante les artéfacts de
mouvement des enfants en comparaison avec le mode hélicoïdal. Au final, le mode
d’acquisition volumique sera choisi pour les scanners avec une faible couverture d’acquisition.
A l’inverse, pour les scanners ayant une grande couverture d’acquisition, le mode hélicoïdal
sera préféré.

Article 2 : Amélioration de la qualité d’image scanographique en utilisant les

reconstructions itératives Adaptive Iterative Dose Reduction avec une acquisition wide-
volume sur un scanner 320-détecteurs.

Notre étude confirme que les reconstructions itératives AIDR permettent de réduire de
manière significative le bruit de l’image par rapport aux reconstructions standard FBP. La
réduction du bruit de l’image était calculée à 31 % sur les scanners lombaires des patients.
Etant donné qu’il existe une relation directe entre le bruit de l’image et la dose (le bruit de
l’image est inversement proportionnel à la racine carrée du mA et le mA est directement
proportionnel à la dose [38]), notre étude nous a permis d’estimer qu’il était possible de
réduire d’environ 52 % la dose de nos scanners lombaires avec les reconstructions itératives
AIDR par rapport aux reconstructions standard en FBP. Par ailleurs, notre étude sur fantôme a
montré qu’avec l’utilisation des reconstructions itératives la réduction du bruit de l’image ne
s’accompagne pas d’une altération de la résolution spatiale. Ce point est crucial car en
pratique clinique d’autres méthodes permettent de réduire le bruit de l’image. En particulier,
l’utilisation d’un filtre plus « mou » permet de lisser l’image, de même que l’épaississement
des coupes. Toutefois, ces deux méthodes permettant de réduire le bruit de l’image sont aussi
à l’origine d’une dégradation de la résolution spatiale, à l’inverse des reconstructions

130
itératives. Par contre, l’utilisation des reconstructions itératives est à l’origine d’une
modification de l’aspect des images qui apparaissent « lissées » ou « informatisées ». Cette
modification de l’aspect des images est principalement liée à la modification de la distribution
des spectres de fréquence spatiale du bruit entre les images FBP et AIDR [41].
Malheureusement, lors de cette étude, il ne nous a pas été techniquement possible de calculer
ces spectres afin de confirmer ces modifications. De même, lors de l’étude sur patient, nous
n’avons pu calculer directement la réduction de la dose car il n’était pas éthique de faire deux
acquisitions (une à dose normale avec les reconstructions FBP et une à dose réduite de moitié
avec les reconstructions AIDR) pour un même patient. Enfin, nous n’avons pu évaluer les
reconstructions AIDR qu’en mode volumique car celles-ci n’étaient pas disponibles pour les
acquisitions hélicoïdales.

Article 3 : Réduction de la dose des scanners abdominopelviens grâce aux reconstructions

itératives AIDR 3D.

Notre étude sur patient évaluant l’impact des reconstructions itératives AIDR 3D a montré
qu’il était possible de réduire de moitié la dose d’un scanner abdominopelvien avec les
reconstructions itératives AIDR 3D par rapport aux reconstructions standard en FBP. Ces
résultats ont été obtenus grâce à l’utilisation de deux scanners d’un même patient réalisés sans
et après implantation des reconstructions itératives. Sous réserve de l’évaluation de l’absence
de modification du morphotype des patients entre les deux scanners, nous avons ainsi pu
montrer directement la réduction de la dose à qualité d’image équivalente grâce aux
reconstructions itératives AIDR 3D. A partir d’une même acquisition, nous avons aussi
montré que l’utilisation des reconstructions itératives par rapport aux reconstructions en FBP
était à l’origine d’une amélioration significative de la qualité d’image objective et subjective.
Cette étude montre bien les différentes manières d’utiliser les reconstructions itératives en
pratique clinique : soit elles permettent de réduire la dose à qualité d’image constante, soit
elles améliorent la qualité d’image à dose constante. Il est aussi possible de réduire un peu la
dose tout en améliorant la qualité d’image. Ces différentes options seront choisies notamment
en fonction du contexte clinique. Par exemple, chez un patient jeune la réduction de la dose
sera favorisée alors que pour une personne âgée pour laquelle le risque de cancer radio-induit
devient négligeable, l’amélioration de la qualité d’image sera plus intéressante. Dans notre
étude, nous avons aussi retrouvé une patiente porteuse d’un dispositif intra-utérin métallique

131
pour laquelle les reconstructions itératives AIDR 3D permettaient de réduire les artéfacts
métalliques et de renforcement du faisceau par rapport aux images FBP. Il semble donc que
ces reconstructions AIDR 3D permettent également de réduire ces artéfacts. Une autre étude
portant sur des scanners de patient avec des prothèses métalliques serait intéressante pour
vérifier cette hypothèse.

Article 4 : Scanner basse dose avec la modulation automatique du milliampérage, les

reconstructions Adaptive Statistical Iterative Reduction et un faible kilovoltage pour le
diagnostic des coliques néphrétiques : impact de l’indice de masse corporelle.

Notre étude a permis de montrer que malgré l’utilisation de la modulation automatique du

mA durant l’acquisition d’un scanner basse dose pour le bilan d’une colique néphrétique, les
scores de qualité d’image subjective et de confiance dans le diagnostic étaient
significativement meilleurs chez les patients avec un IMC > 25 kg/m² par rapport aux patients
avec un IMC < 25 kg/m². Pourtant, la mesure objective du bruit de l’image confirme le bon
fonctionnement de la modulation automatique du mA avec un niveau de bruit de l’image
constant quel que soit le morphotype du patient. Ces résultats peuvent sembler paradoxaux,
mais ils sont liés au fait qu’à un niveau de bruit de l’image équivalent, il est plus facile de
diagnostiquer une colique néphrétique lorsque le patient a beaucoup de graisse intra-
abdominale et intra-pelvienne. En effet, la présence de graisse permet de délimiter plus
facilement le trajet des uretères, même avec un bruit important de l’image. Cet exemple met
en avant la difficulté de trouver des critères pertinents de mesure de la qualité d’image et
l’absence de corrélation entre certains paramètres de mesure de la qualité de l’image et la
performance diagnostique. Au final, dans une démarche de réduction de la dose au scanner,
l’évaluation de la performance diagnostique doit être privilégiée par rapport à l’évaluation de
la qualité d’image. Au cours de cette étude, nous avions aussi trouvé de nombreux scanners
réalisés dans le cadre du bilan d’une colique néphrétique mais pour lesquels le protocole avait
été modifié. En effet, sur 146 scanners réalisés en 2012 pour suspicion de colique néphrétique,
seulement 86 scanners avaient été réalisés selon le protocole basse dose et inclus dans notre
étude. Soixante scanners avaient un indice de bruit diminué, c’est-à-dire que la qualité
d’image avait été améliorée mais aussi que la dose était plus importante. Suite à cette
découverte et après information auprès du personnel du service d’imagerie de l’HIA Legouest,
il s’est avéré que certains manipulateurs et certains radiologues amélioraient volontairement la

132
qualité d’image des scanners qu’ils trouvaient trop dégradée. Pourtant, l’analyse des données
de notre étude a confirmé l’excellente performance diagnostique de notre protocole basse
dose, malgré la dégradation importante de la qualité d’image. Cela montre bien la difficulté de
mettre au point un protocole basse dose avec une qualité d’image dégradée et l’importance de
la sensibilisation des équipes médicales et paramédicales dans une démarche d’optimisation
de la dose.

133
134
 CHAPITRE 4 : APPLICATIONS CLINIQUES

Ce chapitre est composé de trois articles :

1- Gervaise A, Gervaise-Henry C, Pernin M, Naulet P, Junca-Laplace C, Lapierre-

Combes M. Low dose CT for renal colic: how to do in clinical practice? Diagn Interv
Imaging 2016; 97: 393-400.

2- Gervaise A, Teixeira P, Villani N, Lecocq S, Louis M, Blum A. Dose optimization

and reduction in musculoskeletal CT. Diagn Interv Imaging 2013; 94: 371-88.

3- Teixeira P, Gervaise A, Louis M, Lecocq S, Raymond A, Aptel S, Blum A.

Musculoskeletal Wide detector CT : Principles, Techniques and Applications in
Clinical Practice and Research. Eur J Radiol 2015; 84: 892-900.

135
136
 La mise en œuvre pratique des différents facteurs comportementaux et techniques
permettant d’optimiser et de réduire la dose d’irradiation au scanner a déjà été décrite dans de
nombreuses publications [35, 49-55]. Nous proposons de mettre en avant ces modalités de
réduction de la dose dans deux domaines d’imagerie particuliers : l’imagerie des coliques
néphrétiques et l’imagerie ostéo-articulaire.

Article 1 : Scanner basse dose pour la recherche d’une colique néphrétique : comment
faire en pratique clinique ?

Le scanner est devenu l’examen de référence pour l’étude des coliques néphrétiques [56-
57]. Il permet d’en faire le diagnostic, de définir la prise en charge et de rechercher des
diagnostics différentiels. Sa principale limite est liée à son caractère irradiant, d’autant plus
que la maladie lithiasique urinaire touche principalement des sujets jeunes avec une tendance
à la récidive [58-59]. La réduction de la dose des scanners réalisés pour suspicion de colique
néphrétique est donc primordiale.

Le but de cette mise au point était de montrer comment il est possible de réaliser en
pratique clinique courante un scanner basse dose dans le cadre d’une suspicion de colique
néphrétique.

Dans cette mise au point, nous illustrons les différentes modalités de réduction de la dose
de ces scanners en distinguant les facteurs comportementaux et les facteurs techniques. Parmi
les facteurs comportementaux, la réduction de la couverture d’acquisition est un moyen
simple et efficace pour réduire la dose. Les facteurs techniques s’appuient principalement sur
l’utilisation de la modulation automatique du mA, la baisse du kV et du mA et l’implantation
des reconstructions itératives. Grâce à ces mesures d’optimisation de la dose, il est possible de
réaliser des scanners basses doses avec une excellente performance diagnostique et une dose
réduite de l’ordre de 75 % par rapport à un scanner abdominopelvien standard.

Article 2 : Optimisation et réduction de la dose en scanner ostéo-articulaire.

Grâce à ses bonnes résolutions temporelle et spatiale, le scanner reste indiqué dans
l’évaluation de nombreuses pathologies ostéo-articulaires. De nouvelles techniques

137
d’exploration telles que le scanner dynamique 4D des articulations et le scanner de perfusion
tumorale apportent aussi de nouvelles indications. Le scanner reste toutefois une technique
d’imagerie irradiante pour laquelle l’optimisation et la réduction de la dose sont primordiales.

Le but de cette mise au point était de présenter les doses typiques délivrées au cours des
scanners en pathologie ostéo-articulaire et d’illustrer les différentes modalités permettant
d’optimiser et de réduire ces doses en distinguant les facteurs comportementaux et les facteurs
techniques.

L’optimisation du mA et du kV reste indispensable. Ils doivent être adaptés au type

d’exploration et au morphotype de chaque individu. L’apparition récente des algorithmes de
reconstruction itérative a aussi permis de réduire de manière importante les doses délivrées.
Avec l’ensemble de ces techniques d’optimisation et de réduction de la dose, il est dorénavant
possible de faire des acquisitions basses-doses voire très-basses-doses permettant d’atteindre
un niveau de dose parfois proche d’un bilan radiographique standard. Toutefois, même si ces
facteurs techniques permettent de réduire de façon importante les doses délivrées, les facteurs
comportementaux, comme le respect des indications ou la limitation de la couverture
d’acquisition, restent fondamentaux.

Article 3 : Scanner ostéo-articulaire à large système de détection : principes, techniques

et applications en pratique clinique et en recherche.

Le développement des scanners à large système de détection a permis la mise au point de

nouvelles applications avancées en scanner ostéo-articulaire : scanner dynamique 4D des
articulations et scanner de perfusion tumorale. Bien que ces techniques semblent jouer un rôle
important dans le diagnostic de nombreuses pathologies ostéo-articulaires, elles nécessitent la
répétition de multiples volumes d’acquisition, ce qui est à l’origine d’une augmentation
importante des doses délivrées. La maitrise des protocoles d’acquisition et des doses délivrées
est un facteur important pour permettre l’utilisation en pratique clinique de ces nouvelles
applications avancées.

Le but de cette mise au point était de fournir un guide pratique permettant d’utiliser ces
nouveaux outils en routine clinique.

138
 Dans cette mise au point, nous montrons que l’utilisation d’un scanner à large système de
détection permet de réaliser des acquisitions volumiques qui ont pour avantage de supprimer
l’effet d’overranging par rapport à une acquisition hélicoïdale classique et donc de réduire les
doses délivrées. Par ailleurs, l’implantation des reconstructions itératives permet de réduire de
moitié les doses délivrées. En parallèle de ces évolutions technologiques, c’est aussi
l’optimisation des facteurs comportementaux qui reste indispensable. La limitation de la
couverture d’acquisition, la réduction du nombre de phases d’acquisition et l’utilisation d’une
acquisition intermittente plutôt que continue sont les principaux moyens comportementaux
qui permettent de limiter la dose d’irradiation et l’utilisation en pratique clinique courante de
ce type d’applications avancées.

139
140
 Chapitre 4

Article 1 : Scanner basse dose pour la recherche d’une colique néphrétique : comment faire
en pratique clinique ?

Gervaise A, Gervaise-Henry C, Pernin M, Naulet P, Junca-Laplace C, Lapierre-Combes M. Low dose CT for

renal colic: how to do in clinical practice? Diagn Interv Imaging 2016; 97:393-400.

141
142
Diagnostic and Interventional Imaging (2016) 97, 393—400

REVIEW /Genito-urinary imaging

How to perform low-dose computed

tomography for renal colic in clinical
practice
A. Gervaise a,∗, C. Gervaise-Henry b, M. Pernin a,
P. Naulet a, C. Junca-Laplace a, M. Lapierre-Combes a

a
Department of medical imaging, HIA Legouest, 57077 Metz, France
b
Department of biochemistry, hôpital Central, CHU de Nancy, 54000 Nancy, France

KEYWORDS Abstract Computed tomography (CT) has become the reference technique in medical imaging
Computed for renal colic, to diagnose, plan treatment and explore differential diagnosis. Its main limita-
tomography (CT); tion is the radiation dose, especially as urinary stone disease tends to relapse and mainly affects
Dose; young people. It is therefore essential to reduce the CT radiation dose when renal colic is sus-
Optimization; pected. The goal of this review was twofold. First, we wanted to show how to use low-dose CT
Reduction; in patients with suspected renal colic in current clinical practice. Second, we wished to discuss
Renal colic the different ways of reducing CT radiation dose by considering both behavioral and technolog-
ical factors. Among the behavioral factors, limiting the scan coverage area is a straightforward
and effective way to reduce the dose. Improvement of technological factors relies mainly on
using automatic tube current modulation, lowering the tube voltage and current as well using
iterative reconstruction.
© 2015 Éditions françaises de radiologie. Published by Elsevier Masson SAS. All rights reserved.

Since unenhanced (or plain) computed tomography (CT) was introduced in the 1990s, it
has become the reference tool for the diagnosis of renal colic [1—3]. This is because CT
has many advantages. It is fast, does not require intravenous administration of iodinated
contrast material, has high diagnostic capabilities [2,4], helps exclude other conditions
that are clinically similar to renal colic [5—8], provides direct information relative to
the size and attenuation value of urinary stones [9] and helps predict spontaneous stone
passage [10].

∗ Corresponding author at: Department of medical imaging, HIA Legouest, 27, avenue de Plantières, BP 90001, 57077 Metz cedex 3, France.
E-mail address: [Link]@[Link] (A. Gervaise).

[Link]
2211-5684/© 2015 Éditions françaises de radiologie. Published by Elsevier Masson SAS. All rights reserved.
394 A. Gervaise et al.

Its main limitation, however, is the radiation dose given Despite the recent technological advances and the use of
to the patient, especially because urinary stone disease new very powerful iterative algorithms for reconstructions,
tends to relapse and mainly to affect young people. Katz these ultra-low-dose protocols perform less well than low-
et al. report that 4% of the patients that undergo CT for dose protocols for detecting small urinary stones below
suspected renal colic have had at least three CT examina- 3 mm [18,21].
tions for the same indication, with cumulated doses ranging
from 20 to 154 mSv [11]. Considering the ALARA principle
(As Low As Reasonably Achievable) and the potential risks of How to perform low-dose CT to detect
radiation-induced cancer caused even using low doses of X- renal colic?
rays [12,13], dose reduction in CT for suspected renal colic
is hence essential. In this context, many studies have shown The modalities to reduce dose in CT are based on the
that it is possible to detect renal colic with low-dose CT. radioprotection principles of CT dose justification and
Doses may be reduced by 75 to 90% compared to standard optimization [26]. These modalities have already been
acquisition doses, without modifying the diagnostic perfor- extensively described [27—33]. In this review, we discuss
mance [4,14—18]. However, a recent study showed that in them and concentrate on how to reduce the dose of abdomi-
most imaging centers low-dose CT protocols were not used nal and pelvic CT when looking for renal colic. The different
to diagnose renal colic [19]. modalities depend both on behavioral factors, independent
The goal of this review was twofold. First, we wanted of the CT equipment, and technological factors, some of
to show how to use low-dose CT in patient with suspected which depend on how recent the CT equipment is. The
renal colic in current clinical practice. Second we wished to behavioral factors are the level of awareness of the medical
discuss the different ways of reducing the CT radiation dose and paramedical teams, the principles of substitution and
by considering both behavioral and technological factors. justification, as well as limiting the scan coverage area. The
technological factors include reduction of the tube current
and voltage, automatic tube current modulation and iter-
What is low-dose CT? ative reconstructions, as well as optimization of the pitch
and slice thickness.
The definition of low dose is controversial. The term refers
to CT scans where, compared to a ‘‘normal’’ or ‘‘standard’’
Compliance with the indications and
dose scan, the image quality has been deliberately modified
to reduce the exposure dose while preserving the diagnostic substitution with a non-radiating imaging
performance [20]. Renal colic is particularly appropriate for technique
low-dose CT because of the excellent spontaneous contrast
between most urinary stones that are spontaneously hyper- Due to its excellent diagnostic performance, CT has become
attenuating (between 200 and 2800 HU) [2] and the soft the reference investigation to diagnose renal colic. In 2014
tissues that surround them. Thus, even if the dose reduction the European Association of Urology has recommended
is substantial, the naturally high contrast between urinary low-dose CT as the first-line imaging modality in case of
stones and the surrounding soft tissues prevents too much suspected renal colic (grade A recommendation) [34]. In
deterioration of the contrast-to-noise ratio while preserving 2008, the French-speaking Society of medical Emergencies
good diagnostic performance [9]. (Société Francophone d’Urgences Médicales) [35] recom-
Data from the literature reveal that the effective ‘‘low mended radiologists to perform plain abdomen radiography
dose’’ to detect renal colic, is between 1 and 3 mSv [4,19]. together with an ultrasound or an unenhanced CT as a first-
The threshold of 3 mSv (i.e. a dose length product [DLP] line examination for suspected non-complicated renal colic.
of 200 [Link]) is arbitrary but has become the standard However, CT should be favored if a complicated case is sus-
threshold for low-dose CT when investigating renal colic [19] pected or in special situations (pregnancy, single kidney,
because it corresponds more or less to the average radia- transplanted kidney, known uropathy or renal failure) or if
tion of intravenous urography that used to be the reference there are signs of complications (signs of infection; olig-
modality in the past [21]. If we consider that the average uria, anuria or algesia) and in case of doubtful diagnosis.
dose of a standard abdomen and pelvic CT is between 10 In pregnant women, ultrasound must be used as first-line
and 12 mSv [22,23], a low-dose scan of less than 3 mSv cor- modality and, in case of doubtful ultrasound, magnetic res-
responds to a dose reduction of more than 75%. onance imaging should be used as a second-line imaging
Despite this significant dose reduction, various studies modality before CT [36].
have shown that the diagnostic performance of low-dose
CT remains excellent compared to normal-dose CT. A meta- Raising the awareness and training the
analysis published in 2008 showed an average sensitivity of medical teams
96.6% and an average specificity of 94.9% [4]. At the same
time, it was shown that low-dose CT could explore differ- Raising the awareness and training the radiologists and clini-
ential diagnosis, just like normal-dose unenhanced CT [24] cians is also essential [37]. Clinicians must be able to detect
(Fig. 1) and also that there was no significant difference renal colic and ask explicitly the radiologist to look for it.
when determining the size and density of the stones [17,25]. The radiologist must use a low-dose CT protocol with pre-
Recently, experts have suggested using ‘‘ultra-low-dose’’ adjusted parameters. It is also essential that clinicians and
CT, below the level of 1 mSv and close to the dose used to radiologists agree to seek, not the best possible image qual-
perform a plain abdominal radiography, i.e. 0.7 mSv [21]. ity, but one that is sufficient for diagnosis. For radiologists
How to perform low-dose computed tomography for renal colic in clinical practice 395

Figure 1. A 28-year-old woman was admitted to the emergency department for pelvic pain irradiating towards the left lumbar fossa.
Unenhanced abdominal and pelvic CT (100 kVp, noise index at 50, DLP of 74 [Link] and effective dose of 1.1 mSv). Axial views, 1.25 mm
centered on the kidneys (a) and the pelvis (b). Low-dose unenhanced CT does not show any dilatation of the pelvicalyceal system (arrows)
and no wedged urinary stone, thereby excluding the presence of renal colic. However, even if the dose reduction has been significant, it is
possible to evidence intraperitoneal perihepatic effusion (asterisk) as well as a hyperattenuating spontaneous effusion in the Douglas pouch
(arrowhead) suggesting hemoperitoneum. Further enhanced CT confirmed hemoperitoneum caused by left ovarian cyst rupture.

and operators to be properly aware of low-dose CT, they second (Fig. 3) [5,24,43—46]. Many studies have shown
must know the delivered doses. Therefore, it is essential excellent diagnostic performance for low-dose CT, equiv-
that the dose (DLP) be displayed on the CT workstation alent to the one of a standard-dose CT [4]. Hamm et al.
before any acquisition. Currently all manufacturers system- [44] and Poletti et al. [24] have, however, observed that
atically provide this display. Awareness is also raised by low-dose CT performed less well in obese patients who had
the software’s dose-recording system that allows radiolo- a Body Mass Index (BMI) > 30 kg/m2 . This was associated to
gists to monitor the doses absorbed by the patients and the constant mA used for all the patients, resulting in a
to detect cumulated doses, sometimes substantial [38,39]. significant loss of image quality in obese patients. Based
More generally, national and international dose registers are on this, some experts have suggested not using low-dose
available. For instance, the CT Dose Index Registry [40] in CT for obese patients (> 30 kg/m) [24,44,45] while others
the United States has made it possible to evidence that low- have recommended tailoring the mA to these patients [5].
dose CT protocols were not sufficiently used to detect renal After these studies were published, automatic tube cur-
colic [19]. rent modulation during acquisition was introduced. This
has allowed radiologists to adapt the mA and the image
Limiting the scan coverage area quality to the patient’s body mass while reducing the
dose by about 43 to 66% [47,48]. Mulkens et al. con-
A straightforward and effective way to reduce doses is to firmed that low-dose CT with automatic tube current
reduce the acquisition length. Unenhanced image acquisi- modulation provides excellent diagnostic performance in all
tion must be restricted to the urinary tract, from the upper patients with suspected renal colic, including overweight
pole of the kidneys to the base of the urinary bladder. and obese patients [49]. However, in order to preserve
Besides reducing the CT overall dose by limiting the scan an acceptable image quality in overweight patients, the
coverage area, this centering prevents radiosensitive organs automatic tube current modulation increases the CT dose.
such as gonads in men and breasts in women to be exposed Moreover, it has been shown that automatic tube cur-
to X-rays (Fig. 2) [41]. rent modulation provides better scores of image quality
and diagnostic performance for overweight patients with
Reducing the tube current (mA) and tube a BMI ≥ 25 kg/m2 than for patients with a BMI < 25 kg/m2
voltage (kV) [16]. These results may seem inconsistent, but they can
be explained by the fact that, with an equivalent level
Effects of mA and kV of image noise, it is easier to diagnose renal colic in a
patient who has a lot of intra-abdominal and intra-pelvic
Lowering the tube current lowers the dose proportionally fat [50]. Indeed, fat may help delineate the ureters from
but also causes an increase in image noise proportionally surrounding structures, even if the image noise is high. It
to the reciprocal value of the square root of the mA [42]. also seems easier to detect secondary signs of renal colic
In practice, reducing the tube current by half reduces the such as perirenal stranding and the ‘‘rim sign’’ in over-
dose by 50% but increases the image noise by 41%. weight patients. This is why diagnosis errors are more often
Lowering the kV may also reduce the dose. However, this observed in thin patients who have a BMI < 25 kg/m2 , in
will also increase the image noise [42]. whom it is difficult to distinguish small stones in the lower
ureter from pelvic phleboliths, even with normal-dose CT
Effect of patient’s body mass [16,49].
Because of the high natural contrast between most uri- As far as the kV is concerned, beam-hardening artifacts
nary stones and surrounding soft tissues, several experts have been observed in overweight patients if the kV has been
have recommended low-dose CT protocols with significan- too much reduced. So, while it is possible to reduce the tube
tly lowered tube current, by 10 to 100 milliamperes per voltage to 80 kVp in a patient with standard morphotype, it
396 A. Gervaise et al.

Figure 2. 41-year-old woman with suspected left renal colic. Low-dose unenhanced CT followed by standard-dose abdominal and pelvic
enhanced CT (since renal colic was excluded). Scout view (a) shows the borders of the unenhanced (red lines) and enhanced (blue lines)
acquisitions and first and last images in axial view without (b and c) and after injection (d and e). Note the low-dose CT centered from
the upper pole of the kidneys to the mid pubic symphysis making it possible to reduce by 20% the scan coverage area compared to the
standard abdominal and pelvic images (35.1 cm versus 43.7 cm). Also note the presence of mammary tissue (arrow) on the first section of
the standard acquisition, absent in the low-dose series of images.

Figure 3. 30-year-old man monitored for a 4-mm urinary stone in the left kidney (arrow). Normal-dose unenhanced abdominal and pelvic
CT (120 kVp, noise index at 21.4, DLP at 1189 [Link] and effective dose of 17.8 mSv) and (b) follow-up CT with our low-dose protocol
(100 kVp, noise index of 50, DLP of 80 [Link] and effective dose of 1.2 mSv). Even with a 93.5% reduction of the dose, low-dose CT perfectly
shows the left renal stone (arrow).

must be kept to 100 kVp in overweight patients (Figs. 4 and 5) [14—18,52—56]. So, when doses are lowered by mA and
[14,16]. kV reductions, iterative reconstructions compensate for the
decreased image quality. On standard abdominal and pelvic
Iterative reconstructions CT, iterative reconstructions have allowed radiologists to
reduce doses by at least 50% [57]. Kulkarni et al. have shown
Reducing mA and kV is limited by the use of conventional that, for suspected renal colic, it was possible to maintain
Filtered Back Projection (FBP) reconstructions because of excellent diagnostic performance equivalent to the one of
the significant increase in image noise when doses have standard-dose CT by using automatic mA modulation, adap-
been too reduced [51]. The recent introduction of iter- tive statistical iterative reconstruction (ASIR) and a kV fixed
ative reconstruction algorithms has significantly reduced at 80 kVp for patients weighing less than 90 kg [14]. Iterative
image noise compared to standard FBP reconstructions reconstruction also maintains adequate quality of image in
How to perform low-dose computed tomography for renal colic in clinical practice 397

Figure 4. 58-year-old woman (weight, 64 kg; BMI, 22.7 kg/m2 ) with left renal colic caused by a 3.5-mm urinary stone wedged in the
ureterovesical meatus. Unenhanced low-dose CT with tube voltage at 80 kVp and noise index at 50 for a DLP of 105 [Link]. Axial plane,
1.25 mm section (a) and 3-mm (b). Despite the significant reduction of tube voltage, the urinary stone is perfectly visible. However, we
observe beam-hardening artifacts (a) partially reduced by the thickening of the sections (b).

Figure 5. 32-year-old man, obese (BMI, 35.7 kg/m2 ), with suspected right renal colic. Low-dose unenhanced CT (100 kVp, noise index
at 50, DLP of 325 [Link], effective dose of 4.8 mSv) axial view, 1.25-mm sections centered on the kidney (a) and the urinary bladder (b)
and 5-mm coronal MIP reformation (c). For this obese patient, automatic tube current modulation makes it possible to maintain a good
quality of image without the need to increase the CT dose. Note how well it is possible to visualize the infiltration around the right kidney
(asterisk) and the small dilatation on the right side of the pelvicalyceal system (arrow) proximal to a 2-mm urinary stone wedged in the
right ureterovesical meatus (arrow head). The stone is well detected by the 5-mm coronal MIP (c).

overweight patients [16] while using low kV (Fig. 5) and, to reduce image noise while preserving good detectabil-
in addition, has the advantage of reducing beam-hardening ity and characterization of all radiodense urinary stones,
artifacts, including at the pelvis [57]. including those below 3 mm (Fig. 4) [62,63]. Other abdom-
inal structures are also better visualized. However, 5-mm
Pitch effect thickened sections may cause partial-volume artifacts and
reduce the detectability of small stones below 3 mm [64].
Some experts have recommended increasing the pitch in Small stones and spontaneously dense stones are also more
low-dose CT protocols for patients with suspected renal colic readily detected with thickened sections in maximum inten-
[58]. Nowadays, pitch does not affect dose anymore, since sity projection (MIP) and lower image noise. In their study,
most CT have automatic tube current modulation software Corwin et al. have confirmed that urinary stones and their
[59]. However, a high pitch, about 1 to 1.5, is better, because density are more accurately measured on 5-mm coronal MIP
it reduces acquisition time and, thereby, movement artifacts images (Fig. 5) [65].
by the patient.
In routine practice
Adapting the slice thickness
Acquisition must be centered from the upper pole of the
To obtain high spatial resolution, images should always kidneys to the middle of the pubic symphysis. The kV may
be acquired using thin sections (1 to 1.25 mm). Thin be reduced to 100 kVp, even 80 kVp in patients that are not
sections with isotropic voxels enhance the quality of three- overweight, and the level of noise of the automatic tube
dimensional multiplanar reformations and volume rendering current modulation may be increased in order to obtain a
[60]. However, thin sections also cause significant increase 75% reduction of dose compared to a standard abdominal
in image noise, especially if the mA and kV have been con- and pelvic scan protocol. Iterative reconstructions should
siderably reduced, as happens in low-dose CT. So, after be used whenever possible (Table 1). Finally, CT images
using thin sections for image acquisition, it is possible to are visualized on millimetric native axial sections, thick
reconstruct thicker sections during image review at the CT sections (average 3 mm) and 5-mm coronal MIP reforma-
workstation [61]. With thickened 3-mm sections it is possible tions.
398 A. Gervaise et al.

Table 1 Example of a low-dose CT protocol, used in our institution in routine practice to diagnose renal colic with a 64-
slice MDCT (OPTIMA CT660, General Electric Healthcare, USA) and iterative reconstructions (Adaptive Statistical Iterative
Reconstruction [ASIR]). Comparison between acquisition and reconstruction parameters of this low-dose protocol and
those of a standard abdominal and pelvic CT. The differences between the two protocols lie with the limited acquisition
duration, the lowering of the tube voltage and the increase of the noise level of the automatic tube current modulation.
Acquisition and reconstruction Low-dose CT Protocol to detect Standard abdominal and pelvic CT
parameters renal colic Protocol
Acquisition mode/detectors Helical/64 × 0.5 mm Helical/64 × 0.5 mm
Start of acquisition Upper pole of the kidneys Upper border of the diaphragmatic
End of acquisition Middle of the pubic symphysis domes
Lower border of the pubic
symphysis
Tube voltage 80 kVp for a patient with average 120 kVp
BMI
100 kVp for an overweight patient
Tube current (mA) Automatic tube current modulation Automatic tube current modulation
Noise index 50 21.5
Min (mA)/Max (mA) 10/300 120/500
Pitch 1.375 1.375
Rotation time 0.7 0.7
Reconstruction algorithm ASIR 50% ASIR 50%
Slice thickness (mm)/interval (mm) 1.25/1.25 1.25/1.25

Conclusion [3] Roy C, Tuchman C, Guth S, Lang H, Saussine C, Jacqmin D. Scan-

ner hélicoïdal de l’appareil urinaire : principales applications.
CT has become the reference technique to diagnose renal J Radiol 2000;81:1071—81.
colic. Because of its ionizing radiation, it is necessary to [4] Niemann T, Kollmann T, Bongartz G. Diagnostic performance of
reduce doses. In order to perform low-dose CT in patients low-dose CT for detection of urolithiasis: a meta-analysis. AJR
Am J Roentgenol 2008;191:396—401.
with suspected renal colic, the most important measures to
[5] Tack D, Sourtzis S, Delpierre I, de Maertelaer V, Gevenois
implement are: to increase the awareness of the medical
PA. Low-dose unenhanced multidetector CT of patients with
and paramedical teams, to limit the scan coverage area, to suspected renal colic. AJR Am J Roentgenol 2003;180:
use automatic tube current modulation and to reduce mA 305—11.
and kV. Iterative reconstruction algorithms have also made [6] Kluner C, Hein PA, Gralla O, et al. Does ultra-low-dose CT with
it possible to significantly reduce doses (Boxed text 1). Tech- a radiation dose equivalent to that of KUB suffice to detect
nological advances and the introduction of new algorithms renal and ureteral calculi? J Comput Assist Tomogr 2006;30:
for even better iterative reconstructions allow us to expect 44—50.
ultra-low CT with excellent diagnostic performance. [7] Ather MA, Faizullah K, Achakzai I, Siwani R, Irani F. Alter-
nate and incidental diagnoses on noncontrast-enhanced spiral
computed tomography for acute flank pain. Urol J 2009;6:
Boxed text 1: The 5 golden rules of low-dose CT for 14—8.
[8] Samim M, Goss S, Luty S, Weinreb J, Moore C. Incidental find-
suspected renal colic are:
ings on CT for suspected renal colic in emergency department
1. Comply with the indications.
patients: prevalence and types in 5383 consecutive examina-
2. Center and restrict the acquisition coverage area. tions. J Am Coll Radiol 2015;12:63—9.
3. Use automatic tube current modulation. [9] Sung MK, Singh S, Kalra MK. Current status of low dose multi-
4. Lower tube current and tube voltage. detector CT in the urinary tract. World J Radiol 2011;3:
5. Use iterative reconstructions. 256—65.
[10] Coll DM, Varanelli MJ, Smith RC. Relationship of spontaneous
passage of ureteral calculi to stone size and location as
revealed by unenhanced helical CT. AJR Am J Roentgenol
Disclosure of interest 2002;178:101—3.
[11] Katz SI, Saluja S, Brink JA, Forman HP. Radiation dose associ-
The authors declare that they have no competing interest. ated with unenhanced CT for suspected renal colic: impact of
repetitive studies. AJR Am J Roentgenol 2006;186:1120—4.
References [12] Smith-Bindman R, Lipson J, Marcus R, et al. Radiation dose
associated with common computed tomography examinations
[1] Coursey CA, Casalino DD, Remer EM, et al. ACR Appropriateness and the associated lifetime attributable risk of cancer. Arch
Criteria® acute onset flank pain: suspicion of stone disease. Intern Med 2009;169:2078—86.
Ultrasound Q 2012;28:227—33. [13] Berrington de González A, Mahesh M, Kim KP, et al. Projected
[2] Kambadakone AR, Eisner BH, Catalano OA, Sahani DV. New and cancer risks from computed tomography scans performed
evolving concepts in the imaging and management of urolithi- in the United States in 2007. Arch Intern Med 2009;169:
asis: urologists’ perspective. Radiographics 2010;30:603—23. 2071—7.
How to perform low-dose computed tomography for renal colic in clinical practice 399

[14] Kulkarni NM, Uppot RN, Eisner BH, Sahani DV. Radiation dose [33] Gervaise A, Teixeira P, Villani N, Lecocq S, Louis M, Blum A.
reduction at multidetector CT with Adaptive Statistical Itera- CT dose optimization and reduction in osteoarticular disease.
tive Reconstruction for evaluation of urolithiasis: how low can Diagn Interv Imaging 2014;95:47—53.
we go? Radiology 2012;265:158—66. [34] Turk C, Knoll T, Petrik A, et al. Guidelines on urolithiasis.
[15] Winklehner A, Lume I, Winklhofer S, et al. Iterative recons- Arnhem (The Netherlands): European Association of Urol-
tructions versus filtered back-projection for urinary stone ogy (EAU); 2014 [100 p] [Link]
detection in low-dose CT. Acad Radiol 2013;20:1429—35. uploads/22-Urolithiasis [Link].
[16] Gervaise A, Naulet P, Beuret F, et al. Low-dose CT with [35] El Khlebir M, Fougeras O, Le Gall C, et al. Actualisation
automatic tube current modulation, adaptive statistical itera- 2008 de la 8e conférence de consensus de la Société Fran-
tive reconstruction and low tube voltage for the diagnosis of cophone d’Urgences Médicales de 1999. Prise en charge des
renal colic: impact of body mass index. AJR Am J Roentgenol coliques néphrétiques de l’adulte dans les services d’accueil
2014;202:553—60. et d’urgences. Prog Urol 2009;19:462—73.
[17] Wang J, Kang T, Arepalli C, et al. Half-dose non-contrast CT [36] Masselli G, Derme M, Bernieri MG, et al. Stone disease in
in the investigation of urolithiasis: image quality improvement pregnancy: imaging-guided therapy. Insights Imaging 2014;5:
with third-generation integrated circuit CT detectors. Abdom 691—6.
Imaging 2015;40:1255—62. [37] Mayo-Smith WW, Hara AK, Mahesh M, Sahani DV, Pavlicek
[18] Glazer DI, Maturen KE, Cohan RH, et al. Assessment of 1 mSv W. How I do it: managing radiation dose in CT. Radiology
urinary tract stone CT with Model-Based Iterative Reconstruc- 2014;273:657—72.
tion. AJR Am J Roentgenol 2014;203:1230—5. [38] Tamm EP, Szklaruk J, Puthooran L, Stone D, Stevens BL,
[19] Lukasiewicz A, Bhargavan-Chatfield M, Coombs L, Ghita M, Modaro C. Quality initiatives: planning, setting up, and carry-
Weinreb J, Gunabushanam G, et al. Radiation dose index of ing out radiology process improvement projects. Radiographics
renal colic protocol CT studies in the United States: a report 2012;32:1529—42.
from the American College of Radiology National Radiology [39] Burckel LA, Defez D, Chaillot PF, Douek P, Boussel L. Use
Data Registry. Radiology 2014;271:445—51. of an automatic recording system for CT doses: evalua-
[20] Keyzer C, Tack D. Dose optimization and reduction in MDCT tion of the impact of iterative reconstruction on radiation
of the abdomen. In: Tack D, Genevois PA, Kalra M, edit- exposure in clinical pratice. Diagn Interv Imaging 2015;96:
ors. Radiation dose from multidetector CT. Berlin Heidelberg: 265—72.
Springer-Verlag; 2012. p. 369—87. [40] Amis ES. CT radiation dose: trending in the right direction.
[21] McLaughlin PD, Murphy KP, Hayes SA, et al. Non-contrast CT Radiology 2011;261:5—8.
at comparable dose to an abdominal radiograph in patients [41] Corwin MT, Bekele W, Lamba R. Bony landmarks on com-
with acute renal colic; impact of iterative reconstruction on puted tomography localizer radiographs to prescribe a reduced
image quality and diagnostic performance. Insights Imaging scan range in patients undergoing multidetector computed
2014;5:217—30. tomography for suspected urolithiasis. J Comput Assist Tomogr
[22] McCollough CH, Chen GH, Kalender W, et al. Achieving rou- 2014;38:404—7.
tine submillisievert CT scanning: report from the summit on [42] Mahesh M. Scan parameters and image quality in MDCT. In:
management of radiation dose in CT. Radiology 2012;264: Mahesh M, editor. MDCT physics: The basics—–Technology, image
567—80. quality and radiation dose. Philadelphia: Lippincott Williams &
[23] Gervaise A, Esperabe-Vignau F, Naulet P, Pernin M, Portron Y, Wilkins; 2009. p. 47—78.
Lapierre-Combes M. Évaluation des connaissances des prescrip- [43] Liu W, Esler SJ, Kenny BJ, Goh RH, Rainbow AJ, Stevenson GW.
teurs de scanner en matière de radioprotection des patients. Low dose nonenhanced helical CT of renal colic: assessment of
J Radiol 2011;92:681—7. ureteric stone detection and measurement of effective dose
[24] Poletti PA, Platon A, Rutschmann OT, Schmidlin FR, Iselin equivalent. Radiology 2000;215:51—4.
CE, Becker CD. Low-dose versus standard-dose CT protocol [44] Hamm M, Knopfle E, Wartenberg S, Wawroschek F, Wecker-
in patients with clinically suspected renal colic. AJR Am J mann D, Harzmann R. Low dose unenhanced helical computed
Roentgenol 2007;188:927—33. tomography for the evaluation of acute flank pain. J Urol
[25] Sohn W, Clayman RV, Lee JY, Cohen A, Mucksavage P. Low-dose 2002;167:1687—91.
and standard computed tomography scans yield equivalent [45] Heneghan JP, McGuire KA, Leder RA, Delong DM, Yoshizumi
stone measurements. Urology 2013;81:231—4. T, Nelson RC. Helical CT for nephrolithiasis and ureterolithi-
[26] International Commission on Radiological Protection. Recom- asis: comparison of conventional and reduced radiation dose
mendations of the International Commission on Radiological techniques. Radiology 2003;229:575—80.
Protection. ICRP Publication 26. Oxford: Pergamon; 1977. [46] Kim BS, Hwang IK, Choi YW, et al. Low-dose and standard-dose
[27] Semelka RC, Armao DM, Elias J, Huda W. Imaging strategies unenhanced helical computed tomography for the assessment
to reduce the risk of radiation in CT studies, including selec- of acute renal colic: prospective comparative study. Acta
tive substitution with MRI. J Magn Reson Imaging 2007;25: Radiol 2005;46:756—63.
900—9. [47] Kalra MK, Maher MM, D’Souza RV, et al. Detection of urinary
[28] Kalra MK, Maher MM, Toth TL, et al. Strategies for CT radiation tract stones at low-radiation-dose CT with z-axis automatic
dose optimization. Radiology 2004;230:619—28. tube current modulation: phantom and clinical studies. Radi-
[29] McCollough CH, Primak AN, Braun N, Kofler J, Yu L, Christner J. ology 2005;235:523—9.
Strategies for reducing radiation dose in CT. Radiol Clin North [48] McCollough CH, Bruesewitz MR, Kofler Jr JM. CT dose reduction
Am 2009;47:27—40. and dose management tools: overview of available options.
[30] Lee TY, Chhem RK. Impact of new technologies on dose reduc- Radiographics 2006;25:503—12.
tion in CT. Eur J Radiol 2010;76:28—35. [49] Mulkens TH, Daineffe S, De Wijngaert R, et al. Urinary
[31] Singh S, Kalra MK, Thrall JH, Mahesh M. CT radiation dose stone disease: comparison of standard-dose and low-dose with
reduction by modifying primary factors. J Am Coll Radiol 4D MDCT tube current modulation. AJR Am J Roentgenol
2011;8:369—72. 2007;188:553—62.
[32] Dougeni E, Faulkner K, Panayiotakis G. A review of patient dose [50] Katz DS, Venkataramanan N, Napel S, Sommer FG. Can low dose
and optimization methods in adult and paediatric CT scanning. unenhanced MultiDetector CT be used for routine evaluation of
Eur J Radiol 2012;81:e665—83. suspected renal colic? AJR Am J Roentgenol 2003;180:313—5.
400 A. Gervaise et al.

[51] Xu J, Mahesh M, Tsui BM. Is iterative reconstruction ready for suspected renal colic: an effective technique for radiation dose
MDCT? J Am Coll Radiol 2009;6:274—6. reduction? J Comput Assist Tomogr 2000;24:795—801.
[52] Thibault JB, Sauer KD, Bouman CA, Hsieh J. A three- [59] Nagel HD. CT parameters that influence the radiation dose.
dimensional statistical approach to improved image quality for In: Tack D, Genevois PA, editors. Radiation dose from adult
multislice helical CT. Med Phys 2007;34:4526—44. and pediatric multidetector computed tomography. Berlin:
[53] Hara AK, Paden RG, Silva AC, Kujak JL, Lawder HJ, Pavlicek Springer; 2007. p. 51—79.
W. Iterative reconstruction technique for reducing body radi- [60] McNitt-Gray MF. AAPM/RSNA Physics tutorial for resi-
ation dose at CT: feasibility study. AJR Am J Roentgenol dents: topics in CT. Radiation dose in CT. Radiographics
2009;193:764—71. 2002;22:1541—53.
[54] Botsikas D, Stefanelli S, Boudabbous S, Toso S, Becker [61] von Falck C, Galanski M, Shin H. Sliding-thin-slab averaging
CD, Montet X. Model-based iterative reconstruction ver- for improved depiction of low-contrast lesions with radia-
sus adaptive statistical iterative reconstruction in low-dose tion dose savings at thin-section CT. Radiographics 2010;30:
abdominal CT for urolithiasis. AJR Am J Roentgenol 2014;203: 317—26.
336—40. [62] Saw KC, McAteer JA, Monga AG, Chua GT, Lingeman JE, Williams
[55] Greffier J, Fernandez A, Macri F, Freitag C, Metge L, Beregi Jr JC. Helical CT of urinary calculi: effect of stone composi-
JP. Which dose for what image? Iterative reconstruction for CT tion, stone size, and scan collimation. AJR Am J Roentgenol
scan. Diagn Interv Imaging 2013;94:1117—21. 2000;175:329—32.
[56] Greffier J, Macri F, Larbi A, Fernandez A, Khasanova E, Pereira [63] Ketelslegers E, Van Beers BE. Urinary calculi: improved detec-
F, et al. Dose reduction with iterative reconstruction: optimiza- tion and characterization with thin-slice multidetector CT. Eur
tion of CT protocols in clinical practice. Diagn Interv Imaging Radiol 2006;16:161—5.
2015;96:477—86. [64] Memarsadeghi M, Heinz-Peer G, Helbich TH, et al. Unenhanced
[57] Gervaise A, Osemont B, Louis M, Lecocq S, Teixeira P, multi-detector row CT in patients suspected of having urinary
Blum A. Standard dose versus low-dose abdominal and pelvic stones disease: effect of section width on diagnosis. Radiology
CT: comparison between filtered back projection versus 2005;235:530—6.
adaptive iterative dose reduction 3D. Diagn Interv Imaging [65] Corwin MT, Hsu M, McGahan JP, Lamba WM. Unenhanced
2014;95:47—53. MDCT in suspected urolithiasis: improved stone detection
[58] Diel J, Perlmutter S, Venkataramanan N, Mueller R, Lane MJ, and density measurements using coronal Maximum-Intensity-
Katz DS. Unenhanced helical CT using increased pitch for Projection images. AJR Am J Roentgenol 2013;201:1036—40.
 Chapitre 4

Article 2 : Optimisation et réduction de la dose en scanner ostéo-articulaire.

Gervaise A, Teixeira P, Villani N, Lecocq S, Louis M, Blum A. Dose optimization and reduction in
musculoskeletal CT. Diagn Interv Imaging 2013 ; 94 :371-88.

151
152
Diagnostic and Interventional Imaging (2013) 94, 371—388

REVIEW / Musculoskeletal imaging

CT dose optimisation and reduction

in osteoarticular disease
A. Gervaise a,b,∗, P. Teixeira a, N. Villani c,
S. Lecocq a, M. Louis a, A. Blum a

a
Service d’imagerie Guilloz, hôpital central, CHU de Nancy, 29, avenue du
Maréchal-de-Lattre-de-Tassigny, 54035 Nancy cedex, France
b
Service d’imagerie médicale, hôpital d’instruction des armées Legouest, 27, avenue de
Plantières, BP 90001, 57077 Metz cedex 3, France
c
Unité de radiophysique médicale, centre Alexis-Vautrin, avenue de Bourgogne, 54511
Vandœuvre-lès-Nancy, France

KEYWORDS Abstract With an improvement in the temporal and spatial resolution, computed tomography
Dose; (CT) is indicated in the evaluation of a great many osteoarticular diseases. New exploration tech-
Osteoarticular niques such as the dynamic CT and CT bone perfusion also provide new indications. However, CT
imaging; is still an irradiating imaging technique and dose optimisation and reduction remains primordial.
Optimisation; In this paper, the authors first present the typical doses delivered during CT in osteoarticular
Reduction; disease. They then discuss the different ways to optimise and reduce these doses by distin-
CT guishing the behavioural factors from the technical factors. Among the latter, the optimisation
of the milliamps and kilovoltage is indispensable and should be adapted to the type of explo-
ration and the morphotype of each individual. These technical factors also benefit from recent
technological evolutions with the distribution of iterative reconstructions. In this way, the dose
may be divided by two and provide an image of equal quality. With these dose optimisation and
reduction techniques, it is now possible, while maintaining an excellent quality of the image,
to obtain low-dose or even very low-dose acquisitions with a dose sometimes similar that of
a standard X-ray assessment. Nevertheless, although these technical factors provide a major
reduction in the dose delivered, behavioural factors, such as compliance with the indications,
remain fundamental. Finally, the authors describe how to optimise and reduce the dose with
specific applications in musculoskeletal imaging such as the dynamic CT, CT bone perfusion and
dual energy CT.
© 2012 Éditions françaises de radiologie. Published by Elsevier Masson SAS. All rights reserved.

∗ Corresponding author.
E-mail address: [Link]@[Link] (A. Gervaise).

2211-5684/$ — see front matter © 2012 Éditions françaises de radiologie. Published by Elsevier Masson SAS. All rights reserved.
[Link]
372 A. Gervaise et al.

Since its introduction in the 1970s, computer tomogra- CT and the perfusion CT. The dual energy CT is based
phy (CT) has played a major role in the diagnosis of a great on double acquisition with two X-ray beams of different
many osteoarticular diseases. It has quickly become the kilovoltage. This better characterises the tissue and also
choice examination in the diagnosis of traumatic, degen- reduces metallic artefacts or even provides access to bone
erative or even malformative lesions. Even though image subtraction and iodine contrast product [6]. In addition,
quality is altered by metallic artefacts, CT also found indi- the tumour perfusion CT, with the acquisition of successive
cations in postsurgical imaging [1—3]. It is now also used in multiple phases, provides functional information to better
interventional imaging (injection guidance, bone and soft analyse bone and soft tissue tumours. In addition the func-
tissue biopsies, vertebroplasty, etc.) [4]. However, the per- tional analysis is more reproducible than that of the MRI
formance of CT is limited by the inferior analysis of the soft [7,8].
tissue compared with Magnetic Resonance Imaging (MRI). As opposed to the MRI, the other advantages of CT are
The CT analysis of intra-articular lesions is also very difficult represented by its lower cost, improved availability, the pos-
due to the absence of administration of intra-articular con- sibility of use on postsurgical or unstable patients and the
trast product. CT is also a technique of irradiating imaging. absence of contra-indications related to prosthetic materi-
For all of these reasons, the MRI has taken a preponderant als or pacemakers [4,9].
role in musculoskeletal imaging. Finally, CT has benefited from a great many technologi-
Nevertheless, the scanner plays an important role in cal innovations over the last few years, thereby considerably
osteoarticular disease with the development of multislice reducing the dose delivered. The best example is the recent
CT, the development of multidetector CT and recent tech- appearance of iterative reconstructions that reduce the
nological evolutions that reduce the dose the patient is dose by half with an equivalent image quality [10]. With
exposed to. Over the last years, the speed of acquisition and these technological innovations and better control of the
the temporal and spatial resolution of CT have also consid- optimisation of the acquisition parameters, it is now possi-
erably improved. Sub-millimetric isotropic acquisitions are ble to obtain CT imaging with a dose almost equal that of
the rule and multidetector row and three-dimensional (3D) the standard X-ray assessment while the diagnostic perfor-
Volume Rendering (VR) reformations improve the evalua- mance of Ct is much higher than that of X-rays. With the
tion of bone and soft tissue lesions [5]. The improvement continued reduction in the dose delivered, the replacement
in the speed of acquisition reduces movement artefacts and of the X-ray assessment by CT seems to be possible in an
thereby makes the exploration of large volumes possible. increasing number of clinical situations.
This is, for example, especially adapted for the muscu- After a review of the typical doses delivered during
loskeletal assessment of multiple trauma patients (Fig. 1). osteoarticular CT, we will in turn discuss the different meth-
Other technological advances in osteoarticular imaging ods to reduce the dose by emphasising both the behavioural
are represented by the development of the dual energy factors and the technical factors.

Figure 1. Whole-body computed tomography (CT) in a 55-year-old woman for an assessment after falling out of a window. Acquisition
with 64-slice CT covering the whole-body at arterial time, that is, an acquisition of 163 cm in 31 s, with 120 kV, automatic modulation of the
dose with mAs between 50 and 134, rotation time of 0.5 s, 64 × 0.5 mm, pitch at 0.868, for a DLP of 1428 mGy cm. Reformation 3D volume
rendering (VR) of the whole-body (a) then centred on the left femoral fracture (b) and sagittal reformation on the whole vertebral column
(c). This acquisition is obtained in thin slices, providing reformations in 3D VR in order to obtain a global view of the complex fracture of
the left femur and its relationship to the superficial femoral artery and help the surgeon with the presurgical assessment. The fractures of
both patellas should be noted. The multidetector row reformations with bone filter help better analyse the whole vertebral column and
reveal a fracture of the upper vertebral body T11 without recession of the vertebral body.
CT dose optimisation and reduction in osteoarticular disease 373

Typical doses delivered with CT (Table 2). The great variability in doses is mainly due
to the difference in the length of acquisition between the
osteoarticular computer tomography (CT) studies. For example, based on a single-slice CT, Galanski
The International Commission on Radiological Protection and et al. [23] found a mean dose of 2.7 mSv for a mean length
the European Commission recommend the establishment of acquisition of only 5.8 cm. However, with a 16-slice CT,
of diagnostic reference levels (DRL) [11,12]. The European Biswas et al. [22] found a mean dose of 19.15 mSv but for
Commission recommended doses defined by the Weighted- a mean length of acquisition of 25.5 cm. Therefore, more
Dose-CT-Index (CTDI[w]) and by the Dose-Length Product than the difference in terms of number of CT slices, the
(DLP) for several types of CT [12]. For the lumbar vertebrae, dose difference is above all now related to their use. In fact,
the recommended reference levels are a CTDI(w) of 35 mGy while the passage from the single-slice CT to the multislice
and a DLP of 800 mGy cm. For the bony pelvis, (hips, sacroil- CT was accompanied by an increase in the dose delivered
iacs), the recommended reference levels are a CTDI(w) at to the patient [24], the passage from 4-slice CT to 16 or
25 mGy and a DLP at 520 mGy cm. For the traumatic spine, 64-slice CT is accompanied by technical improvements resul-
the recommended values are a CTDI(w) at 70 mGy and a ting in a relatively stable dose [25—27]. Therefore, more
DLP at 460 mGy cm [12]. However, these doses are based than the number of slices, the overall increase in the num-
on reports dating from the end of the 1980s and the begin- ber of CT carried out [28] and the increase in the length
ning of the 1990s, before the introduction of spiral and of acquisition now result in an increase in the individual
multislice CT [13,14]. Since then, the multislice CT radi- and collective exposure [20,29]. Within the same institu-
cally changed practices. In 2004, The European Commission tion, significant variations in terms of ISDP and DLP are also
published new recommendations taking multislice CT into observed (Tables 2—4) [30,31]. This may be accounted for
account. However, they did not recommend new dosime- by the adjustment of the acquisition parameters accord-
try references in terms of DLP in the osteoarticular realm ing to the patient morphotype and the indications. The
[15]. In France, the DRL have recently been up-dated [16]. acquisition parameters may be reduced in the exploration
Among those involving adult computed tomography, only one of the bone structures, while the milliamps increase when
osteoarticular examination is included and only for the lum- as assessment of the soft tissue is required. The establish-
bar vertebrae with a DLP of 700 mGy cm. These reference ment of new techniques of dose reduction, such as iterative
levels are partial and only involve very few explorations in reconstructions, also influences the dose delivered during
osteoarticular imaging. This is all the more so since, with CT (Table 4).
the improvement in the speed of acquisition of multislice Very few studies refer to the peripheral joints. As far
CT, it is now possible to obtain whole spinal imaging, lead- as we are aware, Biswas et al. [22] reported the only
ing to new indications such as the possibility of obtaining a study presenting a full analysis of all of the doses delivered
whole-body CT in a myeloma assessment [17,18] or even the in osteoarticular imaging, including the peripheral joints.
acquisition of a whole spinal column in osteoporosis [19]. In These results show that the farther the anatomic zone is
addition, there is no reference dose for acquisitions of the from the trunk, the more the effective dose is minimal or
peripheral joints or for the new perfusion CT or dynamic CT even negligible, as for example for the wrist (Table 2). This is
applications. because the peripheral joints are smaller, thereby allowing
In the literature, few publications have been devoted for a reduction in the acquisition parameters and providing
to CT doses in the realm of osteoarticular imaging and the shorter lengths of acquisition. However, this is mainly due to
results vary greatly. In a review of the literature dating 2008, the fact that the tissue-weighting factor used in calculating
Mettler et al. [20] find a mean effective dose of 6 mSv for the the effective dose is very small in view of the absence of a
spine CT with values ranging from 1.5 to 10 mSv. In another radiosensitive organ nearby. Table 5 sums up the values of
review of the literature dating 2011 on 19 studies, Pantos the tissue-weighting factors used by Biswas et al. [22] in the
et al. [21] find even greater differences in doses, ranging estimate of the effective dose as a function of the differ-
from 0.8 to 15.7 mSv for a lumbar CT, with a median dose of ent anatomic locations of osteoarticular CT (the effective
5.2 mSv (Table 1). In a 2009 study on the analysis of osteoar- dose (E) in mSv is calculated from the Dose-Length Product
ticular CT doses in their institution, Biswas et al. [22] report (DLP) in mGy cm multiplied by a tissue-weighting factor (k)
of mean dose of 19.15 mSv for the acquisition of a lumbar according to the formula: E = DLP × k).

Table 1 Doses of spinal computed tomography (CT) according to a review of the literature by Pantos et al. [21].
Type of CT CTDI(w)a (mGy) DLPa (mGy cm) Effective dosea (mSv)
Cervical vertebrae 44.3 (5.3—103.2) 324 (56—1275) 2.6 (0.3—7.5)
Dorsal vertebrae NA 253 (66—515) 4.6 (1.0—9.8)
Lumbar vertebrae 30.3 (10.6—59.7) 302 (49—870)b 5.2 (0.8—15.7)
NA: not available; DLP: dose-length product.
a The values are indicated as the median and the extreme values in brackets.
b Note the difference in the dose of lumbar CT between the review of the literature by Pantos et al. [21] and the values provided by

Biswas et al. [22] within their institution (Table 2). This difference is mainly due to an increase in the dose after the passage from the
single-slice CT to the multislice CT (Pantos et al. mainly take studies on single-slice CT into account [21] while Biswas et al. use a 16-slice
CT [22]) as well as the increase in the acquisition lengths that also accompanied the distribution of multislice CT.
374 A. Gervaise et al.

Table 2 Doses of the peripheral joint and spinal computed tomography (CT) according to Biswas et al. [22] (gathered
with a 16-slice CT).
Joints CTDI(w)a (mGy) DLPa (mGy cm) Effective dosea (mSv)
Wrist and hand 14.41 ± 15.52 137 ± 134 0.03 ± 0.03
Elbowb 21.52 ± 23.83 293 ± 311 0.14 ± 0.22
Shoulder 19.49 ± 13.77 316 ± 211 2.06 ± 1.52
Hip 19.83 ± 7.67 422 ± 174 3.09 ± 1.37
Knee 18.39 ± 14.43 356 ± 289 0.16 ± 0.12
Ankle and footc 17.88 ± 13.39 310 ± 210 0.07 ± 0.05
Cervical vertebrae 64.17 ± 29.04 1414 ± 831 4.36 ± 2.03
Dorsal vertebrae 64.39 ± 22.23 2171 ± 805 17.99 ± 6.12
Lumbar vertebrae 66.53 ± 21.56 1701 ± 689 19.15 ± 5.63
DLP: dose-length product.
a The values are indicated as the mean ± standard deviation.
b Only elbow (elbow above the head).
c Unilateral.

Table 3 Osteoarticular computed tomography (CT) doses within our institution with our previous 16-slice CT (Sensation
16, Siemens) [30].
Type of CT CTDI(w)a (mGy) DLPa (mGy cm) Effective dose a
(mSv)
Cervical vertebrae 21 (18.5—45.2) 411 (321—766) 1.3 (1—2.4)
Lumbar vertebrae 32 (23.4—56.4) 782 (399—1527) 8.8 (4.5—17.2)
Pelvic bone 21 (15.6—33.4) 602 (366—1359) 4.4 (2.7—9.9)
Shoulders 25 (23.4—35.0) 332 (253—688) 2.2 (1.6—4.5)
Knee 18 (10.9—31.2) 425 (195—757) 0.2 (0.1—0.3)
DLP: dose-length product.
a The values are indicated as the median and the extreme values in brackets.

Ways to reduce the dose in osteoarticular 97/43 European Community Directive [33] and by the ALARA
(As Low As Reasonably Achievable) principle of precaution.
computer tomography (CT) All of these ways have been extensively detailed in the liter-
The ways to reduce the CT dose are based on the three main ature [9,34—38]. We will discuss them, in turn distinguishing
principles of radioprotection: justification, optimisation and the behavioural from the technical factors and focusing on
substitution [32]. They have been adopted by the Euratom their applications in the realm of osteoarticular CT.

Table 4 Doses with the lumber vertebrae computed tomography (CT) and shoulder arthro-CT within our institution.
Gathered with a 320-slice CT (Aquilion One, Toshiba) before and after implant of AIDR 3D iterative reconstructions
(Adaptive Iterative Dose Reduction 3D, second version of the Toshiba iterative reconstructions).
CTDI(w)a (mGy) DLPa (mGy cm) Effective dosea (mSv)
Lumber vertebrae CT
Before implant of iterative reconstructionsb 40.2 ± 11.4 1094 ± 309 12.32 ± 3.5
With AIDR 3D 25.5 ± 11.9 695 ± 338 7.83 ± 3.8
Shoulder artho-CT
Before implant of iterative reconstructionsb 43.9 ± 15.9 611 ± 259 3.98 ± 1.7
With AIDR 3D 16.1 ± 4.3 205 ± 82 1.34 ± 0.5
DLP: dose-length product.
a The values are indicated as the mean ± standard deviation.
b CT imaging acquired by filtered back projection with QDS (Quantum Denoising System, Toshiba).
CT dose optimisation and reduction in osteoarticular disease 375

created, such as, for example, the CT Dose Index Registry

Table 5 Tissue-weighting factors used to calculate
[42]. Under the initiative of the American College of Radiol-
the effective dose for different anatomic locations in
ogy, this dose register aims at obtaining the CT radiation
osteoarticular disease, calculated according to Biswas
doses from a great many American and foreign radiology
et al. [22].
departments in order to compare the doses and harmonise
Computed tomography (CT) Tissue-weighting practices.
factora (mSv/mGy cm)
Shoulder 6.52 Justification and substitution
Elbowb 0.48
Justification and substitution are also two important ele-
Wrist and hand 0.22
ments in osteoarticular imaging where substitution by
Hip 7.31
non-irradiating imaging techniques such as sonography or
Knee 0.44
MRI is often possible [4,9,43]. For example, Oikarinen et al.
Ankle and footc 0.23
[44], in their study on 30 lumbar CT carried out in patients
Cervical vertebrae 3.08
under the age of 35 years, demonstrated that only seven of
Dorsal vertebrae 8.29
them (23%) were indicated. Of the 23 lumbar CT not indi-
Lumbar vertebrae 11.26
cated, 20 of them would have been able to benefit from
a These factors were calculated from the relationship between an MRI whereas imaging was not indicated for the other
the effective dose and the dose-length product by Biswas et al. three patients. Clarke et al. [45] also demonstrated that 90%
[22]. Note that Biswas et al. calculated these factors using of the lumbar CT could be replaced by an MRI. However,
IMPACT dosimetry software based on the ICRP 60 data [11]. New the MRI is not always possible due to problems of claus-
factors have to be re-calculated to take into account the new
trophobia, non-compatible implants, pacemakers or even
ICRP 103 data.
b Only elbow (elbow above the head). precarious medical conditions [9]. In addition, in certain dis-
c Unilateral. eases, the performance of CT is superior that of MRI [4].
For example, in the study on the spine, the CT has proven
to be more sensitive in the detection of bone changes fol-
lowing an infection [46]. CT is also better than MRI in the
characterisation of certain structures such as gas and calci-
Behavioural factors fications. With its good spatial resolution, CT also provides
a better visualisation of scaphoid fractures [47] or even the
Awareness and education detection of certain osteoid ostomas that remain invisible
First, like in other domains, the education and awareness of in MRI [48]. The angio-CT is sometimes more efficient than
radiologists and radiology technicians is an important ele- the angio-MRI in the assessment of vascular invasion of mus-
ment in the reduction of the dose in osteoarticular disease. culoskeletal tumours [49,50]. In our institution, CT imaging
Wallace et al. [39] have shown that, after doctor educa- is indicated in the following situations: complex fractures,
tion, it was possible to obtain a 29% reduction in the dose fractures with suspicion of vascular impairment, fractures-
of lumbar CT at several institutions. This education also dislocations, initial assessment of musculoskeletal tumours,
emphasises the situations in which the reduction in the dose postsurgical monitoring, bone dysplasia and congenital mal-
is especially important. For example, the patient’s age is a formations, disco-radicular spinal disease and assessment of
major factor since the potential risk of radio-induced cancer joint impairment. In addition, arthro-CT may be carried out
due to low-doses of X-rays decreases with age [40]. Special for almost all joints. It provides a better visualisation of
care is the rule for young subjects. In addition, the anatomic superficial cartilaginous lesions than the MRI. It is also pos-
location of the CT is important. The effective dose of an sible to carry our useful multislice reformations with the CT
acquisition at a distance from radiosensitive organs, as is in a postsurgical assessment [51,52]. CT arthrography of the
the case for the peripheral joints, will be negligible (with wrist also enables a better analysis of ligament lesions than
effective doses sometimes lower than those of a chest X- the MRI or MR arthrography [53] while CT arthrography of
ray!) as opposed to lumbar CT imaging or proximal joint CT the shoulder is a very efficient imaging technique for the
imaging (however, it should be noted that the calculation of detection of SLAP lesions [54].
the effective dose does not take into account the radiosen-
sitivity of tissue related to age). However, the awareness of
Length of exploration and number of acquisitions
radiologists and operators requires knowledge of the doses
delivered. From this point of view, the display of the DLP During CT imaging, the dose can be controlled by reducing
on the CT console before the acquisition is indispensable the number of acquisitions (that is, the number of phases)
and currently systematically available for all manufactur- and the length of the zone explored [55]. The length should
ers. This awareness is also increasingly guaranteed by the be restricted to the zone of interest, previously detected
software used in the gathering and analysis of the doses by the CT topogram(s). As indicated above, this is one of
delivered. This software allows for dosimetric monitoring the main reasons in the differences in doses between the
per patient and detects the cumulate dose, which is some- different examinations. As regards the number of acquisition
times high. The software also includes dosimetric warnings phases, CT often only includes one phase without injection
that help optimise the protocols and help monitor the overall in osteoarticular disease. However, with the development of
reduction in the doses during optimisation [41]. More glob- interventional CT imaging and dynamic and perfusion CT, the
ally, national or international dose registers are also being limitation in the number of phases is of prime importance.
376 A. Gervaise et al.

Position and centering and is therefore relatively greater in case of narrower col-
Exact centering of the anatomic zone for the CT imaging limation. Therefore, the use of a reduced number of slices
at the centre of the ring provides optimum image quality should be avoided with a multislice CT. The use of the volume
and dose delivered. The spatial resolution is actually bet- mode also eliminates pre and postspiral irradiation (or over-
ter at the centre of the ring since more data is obtained ranging), characteristic of the spiral mode [61]. The dose
there than at the periphery [56]. Moreover, good centring of additional irradiation due to pre and postspiral irradia-
is especially required with the use of milliamper automatic tion is higher with an increase in the number of detectors
modulation since this modulation considers that the patient and is also proportionally higher for acquisitions of smaller
is at the centre of the ring [57]. In case of poor centring, length [62] as is the case for acquisitions of the peripheral
the automatic modulation significantly increases the dose joints. With the acquisition of shorter anatomic zones with
[58]. The patient’s position also has an effect on the dose a 16 or 64-slice CT, certain authors recommend the use of
and quality of the image. The volume explored should be the axial and non-spiral mode to eliminate the dose due to
as thin as possible to limit the artefacts of beam harden- overranging [34,63].
ing and reconstruction. This is why the shoulders are placed
at a different height during exploration of the pectoral gir- Kilovoltage
dle (Fig. 2). During the imaging of a leg joint (foot, ankle, The reduction in kilovoltage is the source of a major reduc-
knee), the volume explored should be reduced by raising tion in the dose. However, it is also the cause of an increase
the opposite leg. Similarly, the peripheral joints should be in noise (for example, by maintaining the other parame-
acquired as far as possible from the patient’s trunk in order ters constant, a reduction in kilovoltage from 120 to 80 kV
to reduce the dose received by radiosensitive organs. Biswas reduces the dose delivered by a factor of 2.2 [58] but also
et al. [22] demonstrated that the acquisition of an elbow increases the noise by a factor of 2 [58,64]). In practice, the
along the body compared with a position above the head increase in noise results in a deterioration in the quality of
was responsible for a considerable increase in the effective the image that becomes grained. This appearance is harm-
dose (8.35 versus 0.14 mSv). ful during the analysis of structure with small differences
in density (as is the case for the analysis of soft tissue) in
Technical factors view of an alteration in the contrast to noise ratio. How-
ever, it is not harmful for the analysis of bone structures
Type of computer tomography (CT) acquisition due to a high natural contrast. It is therefore possible to
With the development of multislice CT, the spiral mode has acquire peripheral joints (wrist, knee, ankle, foot) at 100
extensively replaced sequential axial acquisition. However, or even 80 kV (Fig. 3). For example, for a CT of the wrist
the appearance of wide-area detector CT has enabled its with a centred acquisition of 6 cm, with 80 kV and 50 mAs,
return. The 320-slice CT has, with a single rotation, helped the quality of the image is satisfactory for the bone analysis,
acquire a volume of up to 16 cm in length, thereby cover- including a cast immobilisation (Fig. 4). This acquisition pro-
ing most joints (shoulder, wrist, hand, hip, sacroiliac, knee, vides a total DLP total of 20.9 mGy cm, corresponding to an
ankle and foot). The advantage of this type of volume acqui- effective dose of 0.0046 mSv (with a tissue conversion fac-
sition is that it considerably reduces the time of acquisition tor of 0.22 mSv/mGy cm according to Biswas et al. [22]). By
(up to 0.175 s for the acquisition of a volume of 16 cm, with- comparison, this effective dose is only 3.3 times as high as
out shift between the first and last slice) and therefore that of an X-ray assessment comprising five wrist incidences
patient movement artefacts. In addition, this type of vol- (4.6 versus 1.38 mSv) [65] and is less irradiating than a front
ume acquisition reduces the irradiation compared with the chest X-ray (about 0.07 mSv) [66]. For thicker proximal joints
spiral mode. In fact, with wide-area detector CT, the shadow (shoulder, hip, sacroiliac, spine), the kilovoltage should be
phenomenon (or overbeaming) is proportionally smaller than adapted to the patient’s morphotype: 120 kV in a patient
that with 16 or 64-slice CT [59,60]. It should be noted that with a standard morphotype, 100 kV in thin patients while
this shadow phenomenon is independent of the collimation in overweight patients, a kilovoltage at 135—140 kV is some-
times required in order to maintain a satisfactory quality
of image. Given that the iodine attenuation value increases
with a decrease in kilovoltage [67], during arthro-CT of the
proximal joints, it is preferable to use a maximum kilovolt-
age at 120 in order to improve the contrast to noise ratio. For
the same reasons, peripheral arthro-CT (wrist, knee, ankle)
may be carried out at 80 kV (Fig. 5). During vascular or per-
fusion exploration, a reduction in the kV is also possible at
100 or even 80 kV according to the thickness of the anatomic
zone to cover [68]. Certain teams have also proposed low-
dose acquisition protocols at 100 kV for the assessment of
spinal trauma [69], myeloma [70] or even at 80 kV for the
assessment of scoliosis [71] or even osteoporosis [72].
Figure 2. Front topogram before arthro-computed tomography
(CT) of the left shoulder. Note the patient’s position with ascension Milliamps
of the contralateral shoulder, allowing for a reduction in the thick-
ness of the zone to scan as well as a reduction in the hardening A milliamp reduction induces a proportional reduction in the
artefacts of the beam. dose delivered as well as an increase in image noise (the
CT dose optimisation and reduction in osteoarticular disease 377

Figure 3. Computed tomography (CT) of the right knee in a 41-year-old woman for a knee trauma. Acquisition with a 320-slice CT
and volume rendering (VR) with 100 kV, 100 mAs, rotation time of 0.5 s, slice thickness 0.5 mm and 16 cm coverage for a total DLP of
93.5 mGy cm, corresponding to an effective dose of about 0.04 mSv. Axial plane of 0.5 mm passing through the anterior tibial tubercle (ATT)
(a) and Reformation 3D VR (b): non-displaced fracture of the ATT with irradiating secondary articular fracture between the tibial spine and
the medial tibial plateau.

noise value in inversely proportional to the square root of for each patient. The development of the automatic modu-
the milliamps). This may be harmful for the interpretation lation in the milliamps in the three planes allowed for the
of the examinations requiring a good contrast to noise ratio, automation of the adaption of the milliamps to the patient’s
such as for the analysis of disco-radicular disease. In their morphotype [74]. Van Straten et al. [75] demonstrated that
study on lumbar CT, Bohy et al. [73] demonstrated that not this modulation was especially useful in the shoulder and
more than a 35% reduction in milliamps was possible in the pelvic regions where it reduced the effective dose by 11
standard protocol since the diagnostic performance dete- and 17% respectively. Its use is also of interest in adapting
riorated beyond this point. In this study, Bohy et al. [73] the milliamps to the patient’s morphology during the acqui-
used constant milliamps but adapted the body mass index sition of lumbar CT imaging, while maintaining the same

Figure 4. Computed tomography (CT) with cast immobilisation of the right wrist in a 19-year-old man for an assessment after 4 months
of a Schernberg type III scaphoid burst fracture. Acquisition with a 320-slice CT with volume rendering (VR) and a height of 6 cm with 80 kV,
50 mAs, rotation time of 0.5 s, slice thickness 0.5 mm and AIDR 3D iterative reconstruction for a dose-length product (DLP) of 20.9 mGy cm,
corresponding to an effective dose of 0.0046 mSv. Axial plane of 0.5 mm (a) and front reformation of 1.5 mm (b) revealing the persistence
of the fracture and absence of signs of consolidation. Note the good quality of the image in spite of the presence of cast immobilisation and
the major reduction in the acquisition parameters and the DLP.
378 A. Gervaise et al.

is preferred for the reduction of metallic artefacts related

to osteosynthesis materials [79].

Slice thickness
In general, the acquisitions are carried out in thin slices (0.5
to 1 mm), required for the analysis of bone structures and
reconstructed in thicker slices (2 to 5 mm) for the analysis
of soft tissue. The sub-millimetric slices improve the spa-
tial resolution, reduce the effects of partial volume and
allow for multiplane reformations [80]. However, at con-
stant noise, the acquisition in thin slices is the cause of an
increase in the irradiation [81]. If there is an excess reduc-
tion in the milliamps, the acquisition in thin slices leads to a
major increase in the image noise. Therefore, whereas the
acquisition is obtained in sub-millimetric slices, during the
interpretation of the images, the thickening of the slices
helps increase the signal to noise ratio [80] and improve the
analysis of the soft tissue [82,83].

Field collimator
Field collimators are placed at the outlet from the tube
Figure 5. Arthro-computed tomography (CT) of the left ankle in a and help limit the beam of exposure at the field chosen,
41-year-old man referred for an assessment of the persistent pain of thereby allowing for a reduction in the dose. The smallest
the ankle after a serious sprain. Acquisition with a 320-slice CT scan- possible field collimation should be used, in particular with
ner with volume rendering (VR) with 80 kV, 50 mAs, slice thickness
the exploration of the small joints.
of 0.5 mm, rotation time of 0.5 s, adaptive iterative dose reduction
(AIDR) 3D iterative reconstruction, with a total dose-length prod-
uct (DLP) of 23.7 mGy cm, corresponding to an effective dose of Iterative reconstructions
0.005 mSv. Note the excellent analysis of the cartilage in spite of The use of CT iterative reconstructions represents major
the major rerduction in the acquisition parameters and the dose. progress in the reduction of the dose (Table 4). The first
results show that they allow for a reduction of up to 50%
of the dose, while maintaining the same image quality
image quality. Mulkens et al. [76] also demonstrated that [84,85]. Until now, few studies have assessed the value
the use of the automatic milliamp modulation in the three of iterative reconstructions in musculoskeletal imaging. In
planes reduced the dose of a lumbar CT by 37%. Mastora our institution, we carried out a study on 15 lumbar CT
et al. [77] also demonstrated that, during the exploration of acquired with volume mode on a 320-slice CT with Adap-
the thoracic outlet syndrome, the use of automatic milliamp tive Iterative Dose Reduction (AIDR), the first version of
modulation allowed for a 35% reduction in the dose without the iterative reconstruction by Toshiba and by comparing
a loss of image quality. Moreover, certain authors proposed them with standard reconstructions in Filtered Back Projec-
carrying out a low-dose protocol with low milliamps. For tion (FBP). Our results found a mean reduction of 31% in
example, Horger et al. [17] demonstrated that the low-dose the image noise with AIDR compared with the FBP images
acquisition of a whole-body CT was possible in the diagno- [10], without an alteration in the spatial resolution. This
sis of lytic lesions and the assessment of the fracture risk noise reduction corresponds to a potential reduction in the
in patients monitored for multiple myeloma. A collimation dose of 52%. These initial results are promising, especially
of 16 × 1.5 mm was used with 120 kV and between 40 and since the versions of iterative reconstruction are quickly
70 mAS milliamps per second. The effective dose of the CT evolving. In our institution, the new version of AIDR 3D iter-
carried out with 40 mAs was only 1.7 times higher than the ative reconstructions was recently installed and new studies
dose of a standard whole-body X-ray assessment (4.1 mSv are required to assess their impact on the reduction in the
versus 2.4 mSv) [17]. dose and the improvement of image quality (Fig. 6). While
these iterative reconstructions are of particular interest in
reducing the dose of CT imaging requiring a good contrast
Pitch to noise ratio, their performance on examinations where
With certain current multislice CT comprising modern tech- bone analysis is of major importance, such as for example
niques of automatic milliamp modulation, the change in the in the search for a fracture, is not as important. In fact,
pitch does not change the dose since it results in an auto- the high natural contrast of the bone structures allows for
matic adaption of the milliamps [78]. A high pitch, of about low-dose noise acquisitions without this affecting the inter-
1.5, is preferable to reduce the time of acquisition and the pretation in a significant manner [86]. Nevertheless, one of
movement artefacts (for example, during the exploration the benefits of iterative reconstructions is the possibility of
of a multi trauma patient). Nevertheless, the pitch should reducing the artefacts related to FBP reconstructions and,
remain under two in order to maintain the optimum quality in particular, the beam hardening artefacts [87] (Fig. 7).
of the multiplane reformations [78] and avoid the appear- This is of particular interest in the analysis of the soft tissue
ance of spiral artefacts [34]. On the other hand, a low pitch and bone structures in contact with osteosynthesis material.
CT dose optimisation and reduction in osteoarticular disease 379

Figure 6. Lumbar computed tomography (CT) in a 67-year-old woman with a body mass index of 31 kg/m2 . Acquisition from T12 to S2 with
135 kV, automatic milliamp modulation with noise index at 8, rotation time of 0.75 s, spiral 64-slice acquisition of 0.5 mm for a dose-length
product (DLP) of 804 mGy cm. Sagittal reformations of 2 mm with standard reconstruction in filtered back projection (a) and with adaptive
iterative dose reduction (AIDR) 3D iterative reconstructions (b). Note the reduction in image noise with the iterative reconstructions, the
origin of improved visualisation of the disc bulge.

Traditionally, the best visualisation of metal materials 16% reduction in the total dose delivered to the patient.
requires an increase in the acquisition parameters such as However, with an acquisition greater than 300 mm with a 64-
the kilovoltage and milliamps as well as a low pitch and slice CT, the pre- and postspiral irradiation accounts for less
a thin collimation. All of these parameters result in an than 3% of the total dose, irrespective of the pitch [90]. In
increase in the dose [88]. Iterative reconstructions allow for osteoarticular disease, this active collimation is therefore of
a reduction in these artefacts while avoiding an increase in great interest in reducing the dose during acquisition with 16
the dose by the optimisation of other parameters (Fig. 8). or 64-slice CT of the shoulders and hips considering the low
coverage and the proximity of radiosensitive organs (thyroid
Noise reduction filter and gonads).

The improvement in the contrast to noise ratio required

for the analysis of the soft tissue, in particular for disco- In practice
radicular disease, may also be obtained by the use of
noise reduction filters by post-treatment software. These The justification and substitution of examinations by CTg-
filters are applied on already reconstructed images. This raphy remains fundamental: ‘‘a scan that is not carried
allows them to be used with any CT image, including 3D out is one that irradiates the least’’. The limitation of the
reformations. As opposed to the filters used during the coverage of CT imaging is also a simple way to reduce
reconstruction of images, some of these noise reduction fil- the irradiation: ‘‘the smaller the acquisition coverage, the
ters seem able to smooth the image without altering the smaller the dose’’. The volume acquired should be as thin as
spatial resolution. However, studies have to be carried out possible and a small field collimation should be used for the
in order to confirm the value of this new post-treatment exploration of small joints. The kilovoltage and milliamps
hardware. should be adapted to the type of acquisition (reduction of
these acquisition parameters for the exploration of a periph-
eral joint, reduction of the kilovoltage for the acquisition
Active collimation of an arthro-CT), the indication (analysis of the soft tis-
Although not accessible in the choice of parameters, this sue requiring higher milliamps than the analysis of bone
technology is worth describing. Sheilds help reduce pre- and structure) and the morphotype of each individual. The new
postspiral irradiation phenomena by using active collimation techniques to reduce the dose should also be used as soon
in the z-axis at the beginning and end of the spiral [89]. as possible (automatic modulation of the milliamps and the
These shields are of particular interest when the pre- and kilovoltage, iterative reconstructions, active collimation).
postspiral irradiation accounts for a large share of the dose Finally, with dynamic or perfusion acquisition, the dose
of patient irradiation, that is, during short acquisition with a should be reduced by limiting the number of acquisition
16 or 64-slice CT. Christner et al. [90] have shown that with a phases (Boxed text 1).
64-slice CT, for a pitch of 1, a beam collimation of 38.4 mm Nevertheless, even if the reduction of the dose is primor-
and an acquisition length of 15 cm, the shields provide a dial, in particular in young subjects, it should not be carried
380 A. Gervaise et al.

Figure 7. Arthro-computed tomography (CT) of the right shoulder in a 73-year-old man. Acquisition with volume rendering (VR) with a
320-slice CT with 120 kV, 150 mAs, rotation time of 0.75 s, slice thickness of 0.5 mm for a dose-length product (DLP) of 175 mGy cm. Filtered
back projection reconstruction in native axial section of 0.5 mm (a) and frontal reformation in 1.5 mm section (b) and adaptive iterative
dose reduction (AIDR) 3D iterative reconstructions in native axial section of 0.5 mm (c) and in frontal reformation in 1.5 mm section (d). Note
the reduction in image noise as well as the major reduction in hardening artefacts of the beam due to the AIDR 3D iterative reconstructions,
resulting in the improved visualisation of the cartilage of the upper edge of the humerus head.

out at the expense of the quality of the image and especially carried out on the peripheral joints [96,97]. The adapta-
of the diagnostic performance: CT imaging with a reduced tion of the acquisition parameters as well as the application
dose that provides poor image quality and does not allow for of recent methods of dose reduction help maintain a low-
a diagnosis is more harmful than CT imaging with a normal dose delivered to the patient, often lower than that with
dose that allows for a proper diagnosis. traditional acquisition on a conventional CT. Therefore, CT
imaging is a tool of functional analysis improving knowledge
of the joint kinematics and its dysfunctions.
Dynamic computed tomography (CT) The dynamic study of joints is possible in spiral mode
imaging of the joints with a 64-slice CT. Tay et al. [98] have shown, in an exper-
imental study, that it is possible to obtain the dynamic
Joint kinematics may be examined by a static study in differ- acquisition of a wrist in four phases with a very low pitch
ent positions or by a continuous dynamic study. The latter (0.1) by using a protocol with retrospective synchronisa-
should be privileged during study of the joint kinematics tion of the movement. However, this technique induces a
[91—93] since the constraints differ between a system in great many movement and ‘‘step’’ artefacts and a major
action and a static system [94,95]. The improvement in the increase in irradiation [98] rendering the efficacy much
temporal resolution of multislice CT and the development lower than that of volume acquisitions with multidetector
of multidetector CT now allows for dynamic studies to be CT.
CT dose optimisation and reduction in osteoarticular disease 381

diagnostic information in several clinical indications: occult

Boxed text 1 The seven gold rules to reduce instabilities of the wrist, femoro-patellar syndrome, pos-
radiation by osteoarticular computed tomography terior conflict of the ankle, thoracic outlet syndrome.
(CT). Nevertheless, these indications are new and an assessment
• Comply with the indications. of the diagnostic performance of this type of examination
• Whenever possible, prefer a non-irradiating method is required in order to justify their use, especially in view
of their irradiating nature. In fact, these dynamic stud-
of imaging.
• Limit the CT coverage. ies require the repetition of several acquisitions leading to
• Limit the number of acquisition phases during an increase in the irradiation when compared with a sin-
gle acquisition. Nevertheless, on a peripheral joint, with
dynamic and perfusion examinations.
• Adapt the kilovoltage and milliamps to the the possibility of obtaining low-dose acquisitions without
disturbing the interpretation of the movement [99], it is
indications and morphology of each patient.
• Reduce the kilovoltage during the exploration of the possible to obtain dynamic acquisitions with an effective
dose under 1 mSv. For example, for the study of the kine-
peripheral joints and arthro-CT.
• Use modern methods to reduce the dose (iterative matics of radio-ulnar deviation of the wrist with 12-phase
volume rendering (80 kV, 11 mAs, rotation time of 0.35 s,
reconstructions, automatic modulation of the
acquisition coverage of 6 cm), the DLP is only 91.2 mGy cm,
milliamps. . .).
corresponding to an effective dose of 0.02 mSv (Fig. 9). With
this low effective dose, it is possible to study several types
of movement (for example for the wrist: flexion/extension,
In our institution, we study joint kinematics with a 320- tightening, ulnar and radial deviation) while maintaining a
slice CT. This allows for the acquisition of a volume up total effective dose much lower than 1 mSv. In the same
to 16 cm long. A rotation time of 0.35 s combined with a way, for a dynamic exploration of the ankle, the DLP is
technique of partial data reconstruction (reconstruction of under 200 mGy cm, corresponding to an effective dose of
data over 180◦ by rotation) provides a temporal resolution 0.046 mSv.
of 0.175 s per rotation. Volume acquisition also has several For dynamic explorations of the hips and shoulders,
advantages: reduction of the dose compared with the spiral it is especially important to reduce the dose due to the
mode by elimination of pre- and postspiral irradiation [61] optimisation of the acquisition parameters. If the dynamic
and acquisition of the entire volume at one time, without exploration only involves bone segments, the high natural
shift between the first and last slices or shift due to the contrast of the bone allows for a considerable reduction in
movement of the table. With this technique, it is possible the kilovoltage and milliamps [100]. It is also important to
to study the kinematics of the peripheral joints and provide reduce and centre the acquisition coverage of the zone of
interest. In addition, even if Hristrova et al. [96] demon-
strated that image quality is improved by the continuous
acquisition of data, the additional irradiation makes acqui-
sition in intermittent mode preferable, with a number of
phases in general limited to 12. On the pelvis, this allows
the dose of irradiation to be kept at less than 10 mSv, cor-
responding to a standard multiphase abdominal and pelvic
examination.

Perfusion computed tomography (CT)

Perfusion CT was first described a great many years ago
[101]. Like with dynamic examinations, the CT perfusion of
musculoskeletal tumours is possible due to an improvement
in the temporal resolution of multislice CT imaging and the
development of multidetector CT. The study of perfusion by
CT combines the advantages of perfusion imaging, providing
data similar to that provided by the MRI but with bet-
ter visualisation of the bone invasion (periosteal reaction,
cortical break, osteolysis), the femoral necrosis and the
neovascularisation. The quantification of perfusion curves
by CT is as easy as with MRI [102]. This tumoral perfusion
Figure 8. Computed tomography (CT) of the right hip in a 70- may be carried out in spiral mode with a multislice CT due
year-old man for an assessment of right coxalgia on a total hip to a 2D scanning movement [103] or in ‘‘step-and-shoot’’
replacement. Spiral acquisition with a 320-slice CT with 135 kV,
volume rendering with a multidetector CT. The advantage
15 mAs, rotation time at 0.5 s, 64 × 0.5 mm and AIDR 3D itera-
of tumoral perfusion with volume rendering is that it is
tive reconstruction for a dose-length product (DLP) of 326 mGy cm.
Frontal reformation in 2 mm section. Note the excellent quality of obtained without moving the table and therefore there are
the image due to a reduction in the metallic artefacts by iterative fewer movement artefacts, helping improve the quality of
reconstructions that also allow for a reduction in the dose. the reconstructions and perfusion curves. This technique
382 A. Gervaise et al.

Figure 9. Dynamic arthro-computed tomography (CT) of the left wrist in a 37-year-old man for an assessment of persistent pain at the
wrist following a trauma. Acquisition with a 320-slice CT with static acquisition of the arthro-CT with volume rendering (VR) with 80 kV,
50 mAs, rotation time at 0.5 s, thickness of the section 0.5 mm, 8 cm coverage for a dose-length product (DLP) of 25.7 mGy cm, then dynamic
acquisition during a movement of radio-ulnar deviation comprising 12 volume acquisitions with 80 kV, 11 mAs, rotation time of 0.35 s, 0.5 mm
section thickness and 6 cm coverage for a total DLP of 91.2 mGy cm. Frontal reformation of the arthro-CT in 1.5 mm section (a) revealing
a transfixing rupture of the lunotriquetral ligament. The 3D VR reformations of the dynamic CT (a: radial deviation; b: neutral position; c:
ulnar deviation) do not reveal instability of the carpus with dynamic movements of the radio-ulnar deviation of the wrist.
CT dose optimisation and reduction in osteoarticular disease 383

Figure 10. Bone perfusion computed tomography (CT) in an 18-year-old man for an assessment of osteoid osteoma of the distal femoral
metaphysis of the left knee. Volume acquisition with a 320-slice CT of 15 phases (acquisition without injection then acquisition after injection
with nine phases every 5 s then five phases every 10 s) with 100 kV, 50 mAs, rotation time of 0,5 s, section thickness of 0.5 mm and 4 cm
coverage for a total dose-length product (DLP) total of 123 mGy cm. 0.5 mm axial sections in filtered back projection (a), with AIDR iterative
reconstruction (b) and after temporal fusion of the different phases (c), 0.5 mm axial sections at arterial time (phase 4) without (d) and
with bone subtraction (e) and perfusion curve (f). Note the deterioration in the quality of the native images (a) due to the major reduction
in the acquisition parameters and the improvement in image quality due to iterative reconstructions (b) and the temporal fusion technique
(c). Also note the hypervascularisation of the nidus of the osteoid osteoma with intense enhancement at arterial time, fully visible due to
the reconstructed images with bone subtraction (d). The hypervascular aspect of the enhancement of the nidus is also confirmed by the
perfusion curve (f) of the nidus (green curve) when compared with the perfusion curve of a region of interest placed in the popliteal artery
(purple curve).
384 A. Gervaise et al.

Figure 10. ( continued ).

also allows for use of the fist acquisition as a bone subtrac- Dual energy computed tomography (CT)
tion mask, improving the detection and characterisation of
bone anomalies. However, these perfusion studies give rise Dual energy CT is based on the acquisition of two super-
to a high increase in the dose delivered [103]. The protocol imposable images with two different kilovoltages. Based on
is optimised by reducing the coverage of the CT imaging, these native images, it is possible to reconstruct a virtual
by limiting the number of acquisition phases and by adapt- image corresponding to any voltage of the x tube [6]. Each
ing the acquisition parameters (reduction in the kilovoltage manufacturer proposes dual energy acquisitions on their CT.
and milliamps). However, this reduction in the acquisition However, the techniques used often differ. This accounts for
parameters leads to the deterioration of image quality. This the differences in terms of performance and clinical applica-
may be compensated by carrying out a temporal fusion of tions between these techniques. For Siemens, dual energy
the different phases. This technique allows for a summation acquisition is obtained from a bi-tube CT. With each rota-
of several images derived from different acquisition phases tion, this allows for an image to be obtained with one tube
in order to obtain an image with less noise and of better set at 80 kV and the other at 140 kV. For the other manu-
quality (Fig. 10). facturers, only one tube is used to create the dual energy.
In our institution, we have, for example, studied the General Electrics uses a generator that allows for a switch
value of tumoral perfusion in the diagnosis and monitor- in 0.5 ms between high and low voltage. It is thereby possi-
ing of osteoid osteomas [104]. In this disease, the MRI may ble, during a single rotation, to switch 500 times and obtain
be faulty [48] and the use of CT imaging more easily indi- two series of raw data in order to reconstruct two images,
cates the diagnosis by detecting the bone reaction around a one at 80 kV and the other at 140 kV. As for Toshiba, it
small nidus. In addition to this anatomic data, the perfusion benefits from the wide coverage of its system of detection
CT detects the hypervascularisation of the nidus (Fig. 10). (16 cm per revolution) to propose two consecutive revolu-
The subtraction of the phases after injection with the first tions with a change in voltage between both revolutions.
acquisition without injection allows for a sequence of bone Finally, Philips uses a double layer of detector, the first
subtraction revealing the bone marrow oedema around the measuring all of the rays transmitted and the second only
nidus. All of this information, usually provided by the MRI, is measuring the hardest beams [105]. Several applications of
now accessible with CT imaging. Nevertheless, the value and the dual energy CT in the osteoarticular domain are now clin-
role as compared with the MRI has to be assessed, since the ically available although still in the assessment phase [6]:
indications for perfusion CT remain limited to when there is improvement in tissue characterisation, bone subtraction,
a doubt as to the diagnosis in view of the irradiant nature differentiation of bone and iodine contrast product or even
of this technique and the youth of the patients monitored. reduction of metallic artefacts. The improvement in tissue
To control the dose of irradiation, we target the zone of CT characterisation was first used for the detection and char-
coverage at the zone of interest (only 4 to 8 cm suffices). acterisation of urate deposits in gout [106,107]. An initial
Moreover, we choose a kV and mAs adapted to the mor- study by Nicolaou et al. [108] demonstrated that the dual
photype of the patient and the anatomic zone. The number energy acquisition of all peripheral joints (elbows, wrists,
of phases is also limited to 15 or 16, with an acquisition hands, knees, ankles and feet) provides good sensitivity
every 5 s pour for the first nine phases (arterial phases) and and good specificity in the detection of locations of topa-
then every 10 s. All of these measures allow for a perfusion ceous gout while the total effective dose ranged from only
CT with a total DLP generally between 100 and 500 mGy cm 2 to 3 mSv. Another dual energy application is the possibility
(Figs. 10 and 11). of obtaining reconstructions in bone subtraction. It is then
CT dose optimisation and reduction in osteoarticular disease 385

Figure 11. Perfusion computed tomography (CT) of a 68-year-old man presenting a glomic tumour of the left thumb. Volume acquisition
with a 320-slice CT and 17 phases (acquisition without injection and then acquisition after injection with 10 phases every 5 s then six phases
every 10 s) with 80 kV, 40 mAs, rotation time of 0.5 s, section thickness of 0.5 mm and 4 cm cover for a total dose-length product (DLP)
of 274 mGy cm. Sagittal reformations in 1.5 mm section after injection (phase 7) without (a) and with bone subtraction (b), 3D volume
rendering (VR) reformation with bone subtraction (c) and map of tumoral perfusion of the blood volume (d). Note the excellent visualisation
of the glomic tumour due to the images in bone subtraction (b) as well as the high quality of the 3D VR reformations revealing the distal
vascularisation of the thumb (c).

possible to detect the bone marrow oedema. Pache et al. Conclusion

[109] demonstrated that it was possible to detect post-
traumatic bone marrow oedema in knee CT imaging, with an Computed tomography is an imaging technique that contin-
increase in irradiation of about 28% compared with a con- ues to benefit from a great deal of technological progress.
ventional CT. The dual energy bone CT may also become These technological developments, in particular with the
an alternative in case of counter-indication for an MRI or if development of iterative reconstructions, provide a consid-
it is unavailable. However, clinical studies are required to erable reduction in the dose delivered to the patient. They
specify the role of this technique compared with the MRI. also make it possible to access new applications such as per-
Subhas et al. [67] demonstrated that the use of dual energy fusion CT or dynamic CT. The modern techniques for dose
on a shoulder arthro-CT provides a better signal to noise reduction are of special interest in applications that involve
ratio with an equivalent dose. Finally, with a dose equivalent repeated acquisition phases. Nevertheless, the optimisation
that of a standard single energy acquisition, Bamberg et al. of the milliamps and kilovoltage, the limitation in the cover-
[110] demonstrated that the use of dual energy allowed for age of CT as well as compliance with the indications are still
a reduction in prosthesis-related metallic artefacts. the main ways to limit the doses delivered to patients. These
386 A. Gervaise et al.

new indications and the possibility of obtaining low-dose or [16] Arrêté du 24 octobre 2011 relatif aux niveaux de références
even very low-dose acquisitions, while maintaining excel- diagnostiques en radiologie et en médecine nucléaire. Journal
lent image quality, give CT a major role in musculoskeletal officiel de la république française, 14 janvier 2012.
disease. While CT has already replaced the standard X-ray [17] Horger M, Claussen CD, Bross-Bach U, Vonthein R, Trabold T,
assessment in certain indications, its role with respect to Heuschmid M, et al. Whole-body low-dose multidetector row
CT in the diagnosis of multiple myeloma: an alternative to
non-irradiating imaging techniques, such as the MRI, still has
conventional radiography. Eur J Radiol 2005;54:289—97.
to be defined. [18] Gleeson TG, Moriarty J, Shortt CP, Gleeson JP, Fitzpatrick P,
Byrne B, et al. Accuracy of whole-body low-dose multide-
tector CT (WBLDCT) versus skeletal survey in the detection
Disclosure of interest of myelomatous lesions, and correlation of disease dis-
tribution with whole-body MRI (WBMRI). Skeletal Radiol
The authors declare that they have no conflicts of interest 2009;38:225—36.
concerning this article. [19] Damilakis J, Adams JE, Guglielmi G, Link TM. Radiation expo-
sure in X-ray-based imaging techniques used in osteoporosis.
Eur Radiol 2001;20:2707—14.
References [20] Mettler FA, Huda W, Yoshizumi TT, Mahesh M. Effective doses
in radiology and diagnostic nuclear medicine: a catalog. Radi-
[1] Blum A, Walter F, Ludig T, Zhu X, Roland J. Scanners multi- ology 2008;248:254—63.
coupes : principes et nouvelles applications scanographiques. [21] Pantos I, Thalassinou S, Argentos S, Kelekis NL, Panayiotakis
J Radiol 2000;81:1597—614. G, Efstathopoulos EP. Adult patient radiation doses from non-
[2] Cotten A, Iochum S, Blum A. 3D imaging in musculoskeletal cardiac CT examinations: a review of published results. Br J
system. In: Baert AL, Caramella D, Bartolozzi C, editors. 3D Radiol 2011;84:293—303.
image processing. Techniques and clinical applications. Berlin [22] Biswas D, Bible JE, Bohan M, Simpson AK, Whang PG, Grauer
Heidelberg, New York: Springer; 2002. p. 247—55. JN. Radiation exposure from musculoskeletal computerized
[3] Fayad LM, Bluemke DA, Fishman EK. Musculoskeletal imaging tomographic scans. J Bone Joint Surg Am 2009;91:1882—9.
with computed tomography and magnetic resonance imaging: [23] Galanski M, Nagel HD, Stamm G. CT radiation exposure risk in
when is computed tomography the study of choice? Curr Probl Germany. Rofo 2001;173:R1—66.
Diagn Radiol 2005;34:220—37. [24] Thomton FJ, Paulson EK, Yoshizumi TT, Frush DP, Nelson RC.
[4] West ATH, Marshall TJ, Bearcroft PW. CT of the muscu- Single versus multidetector row CT: comparison of radiation
loskeletal system: what is left is the days of MRI? Eur Radiol doses and dose profiles. Acad Radiol 2003;10:379—85.
2009;19:152—64. [25] Mori S, Endo M, Nishizawa K, Murase K, Fujiwara H, Tanada S.
[5] Iochum S, Ludig T, Walter F, Fuchs A, Henrot P, Blum A. Intérêt Comparison of patient doses in 256-slice CT and 16-slice CT
de la technique de rendu volumique en pathologie ostéoartic- scanners. Br J Radiol 2006;79:56—61.
ulaire. J Radiol 2001;82:221—30. [26] Jaffe TA, Yoshizumi TT, Toncheva G, Anderson-Evans C, Lowry
[6] Karcaaltincaba M, Aktas A. Dual energy CT revisited with mul- C, Miller CM, et al. Radiation dose for body CT protocols:
tidetector CT: review of principles and clinical applications. variability of scanners at one institution. AJR Am J Roentgenol
Diagn Interv Radiol 2011;17:181—94. 2009;193:1141—7.
[7] Oldrini G, Lombard V, Roch D, Detreille R, Lecocq S, Louis M, [27] Fuji K, Aoyama T, Yamauchi-Kawaura C, Koyama S, Yamauchi
et al. Courbes de rehaussement des tumeurs osseuses et des M, Ko S, et al. Radiation dose evaluation in 64-slice CT exami-
parties molles : comparaison entre scanner et IRM (abstract). nations with adult and paediatric anthropomorphic phantoms.
J Radiol 2009;90:1578. Br J Radiol 2009;82:1010—8.
[8] Goh V, Padhani AR. Imaging tumor angiogenesis: func- [28] Brenner DJ, Hall EJ. Computed tomography an increasing
tional assessment using MDCT or MRI? Abdom Imaging source of radiation exposure. N Engl J Med 2007;357:2277—84.
2006;31:194—9. [29] Richards PJ, George J, Metelko M, Brown M. Spine com-
[9] Semelka RC, Armao DM, Elias J, Huda W. Imaging strategies to puted tomography doses and cancer induction. Spine
reduce the risk of radiation in CT studies, including selective 2010;35:430—3.
substitution with MRI. J Magn Reson Imaging 2007;25:900—9. [30] Blum A, Noel A, Winninger D, Batch T, Ludig T, Ferquel G,
[10] Gervaise A, Osemont B, Lecocq S, Noel A, Micard E, Fel- et al. Dose optimization and reduction in CT of the muscu-
blinger J, et al. CT image quality improvement using adaptive loskeletal system including the spine. In: Tack D, Genevois
iterative dose reduction with wide-volume acquisition on 320- PA, editors. Radiation dose from adult and pediatric multi-
detector CT. Eur Radiol 2012;22:295—301. detector computed tomography. Berlin: Springer; 2007. p.
[11] International Commission on Radiological Protection. Recom- 185—94.
mendations of the International Commission on Radiological [31] Gervaise A, Teixeira P, Villani N, Lecocq S, Louis M, Blum
Protection. ICRP Publication 60. Oxford: Pergamon; 1991. A. Dose optimization and reduction in musculoskeletal CT
[12] European Commission. European guidelines on quality criteria including the spine. In: Tack D, Kalra M, Genevois PA,
for computed tomography. Report EUR 16262. Luxembourg, editors. Radiation dose from adult and pediatric multide-
1999. tector computed tomography. 2nd ed Berlin: Springer; 2012,
[13] Shrimpton PC, Edyvean S. CT scanner dosimetry. Br J Radiol in press.
1998;71:1—3. [32] International Commission on Radiological Protection. Rec-
[14] Hidajat N, Wolf M, Nunnemann A, Liersch P, Gebauer B, Teich- ommendations of the International Commission on Radio-
gräber U, et al. Survey of conventional and spiral CT doses. logical Protection. ICRP Publication 26. Oxford: Pergamon;
Radiology 2001;218:395—401. 1977.
[15] Bongartz G, Golding SJ, Jurik AG, Leonardi M, van Persijn [33] European Community. Council Directive 97/43/EURATOM, 30
van Meerten M, Rodríguez R, et al. European Guidelines for June 1997, on health protection of individuals against the
Multislice Computed Tomography. Funded by the European dangers of ionizing radiation in relation to medical expo-
Commission (Contract number FIGMCT2000-20078-CT-TIP). sure (repealing Directive 84/466/Euratom). Off J Eur Commun
2004. 1997;L18040:22—7.
CT dose optimisation and reduction in osteoarticular disease 387

[34] Kalra MK, Maher MM, Toth TL, Hamberg LM, Blake MA, Shep- reliability of multidetector CT arthrography CT for diagnosis.
ard JA, et al. Strategies for CT radiation dose optimization. Radiology 2011;260:207—15.
Radiology 2004;230:619—28. [55] Rehani MM, Bongartz G, Kalender W. Managing X-ray dose in
[35] McCollough CH, Primak AN, Braun N, Kofler J, Yu L, Christner computed tomography: ICRP special task force report. Ann
J. Strategies for reducing radiation dose in CT. Radiol Clin ICRP 2000;30:7—45.
North Am 2009;47:27—40. [56] Kalra MK, Toth TL. Patient centering in MDCT: dose effects.
[36] Lee TY, Chhem RK. Impact of new technologies on dose reduc- In: Tack D, Genevois PA, editors. Radiation dose from adult
tion in CT. Eur J Radiol 2010;76:28—35. and pediatric multidetector computed tomography. Berlin:
[37] Singh S, Kalra MK, Thrall JH, Mahesh M. CT radiation dose Springer; 2007. p. 129—32.
reduction by modifying primary factors. J Am Coll Radiol [57] Li J, Udayasankar UK, Toth TL, Seamans J, Small WC, Kalra
2011;8:369—72. MK. Automatic patient centering for MDCT: effect on radiation
[38] Dougeni E, Faulkner K, Panayiotakis G. A review of patient dose. AJR Am J Roentgenol 2007;188:547—52.
dose and optimization methods in adult and paediatric CT [58] Mahesh M. Scan parameters and image quality in MDCT. In:
scanning. Eur J Radiol 2012;81:665—83. Mahesh M, editor. MDCT physics: the basics — Technology,
[39] Wallace AB, Goergen SK, Schick D, Soblusky T, Jolley D. Mul- image quality and radiation dose. Philadelphia: Lippincott
tidetector CT dose: clincal pratice improvement strategies Williams & Wilkins; 2009. p. 47—78.
from a successful optimization program. J Am Coll Radiol [59] Perisinakis K, Papadakis AE, Damilakis J. The effect of X-ray
2010;7:614—24. beam quality and geometry on radiation utilization efficiency
[40] Brenner DJ, Shuryak I, Einstein AJ. Impact of reduced patient in multidetector CT imaging. Med Phys 2009;36:1258—66.
life expectancy on potential cancer risks from radiologic [60] Mori S, Nishizawa K, Kondo C, Ohno M, Akahane K, Endo M.
imaging. Radiology 2011;261:193—8. Effective doses in subjects undergoing computed tomogra-
[41] Tamm EP, Rong XJ, Cody DD, Ernst RD, Fitzgerald NE, Kun- phy cardiac imaging with the 256-multislice CT scanner. Eur
dra V. Quality initiatives: CT radiation dose reduction: how J Radiol 2008;65:442—8.
to implement change without sacrificing diagnostic quality. [61] Gervaise A, Louis M, Batch T, Lœuille D, Noel A, Guillemin F,
Radiographics 2011;31:1823—32. et al. Réduction de dose dans l’exploration du rachis lombaire
[42] Amis ES. CT radiation dose: trending in the right direction. grâce au scanner 320-détecteurs : étude initiale. J Radiol
Radiology 2011;261:5—8. 2010;91:779—85.
[43] Borgen L, Ostense H, Stranden E, Olerud HM, Gudmundsen TE. [62] van der Molen AJ, Geleijns J. Overranging in multisection CT:
Shift in imaging modalities of the spine through 25 years and quantification and relative contribution to dose - comparison
its impact on patient ionizing radiation doses. Eur J Radiol of four 16-section CT scanners. Radiology 2007;242:208—16.
2006;60:115—9. [63] Schilham A, van der Molen AJ, Prokop M, Jong HW. Overran-
[44] Oikarinen H, Meriläinen S, Pääkkö E, Karttunen A, Niemi- ging at multisection CT: an underestimated source of excess
nen MT, Tervonen O. Unjustified CT examinations in young radiation exposure. Radiographics 2010;30:1057—67.
patients. Eur Radiol 2009;19:1161—5. [64] Kalender WA, Deak P, Kellermeier M, van Straten M, Vollmar
[45] Clarke JC, Cranley K, Kelly BE, Bell K, Smith PH. Provision of SV. Application and patient size-dependent optimization of
MRI can significantly reduce CT collective dose. Br J Radiol X-ray spectra for CT. Med Phys 2009;36:993—1007.
2001;74:926—31. [65] Noël A, Ottenin MA, Germain C, Soler M, Villani N, Grosprêtre
[46] Tins BJ, Cassar-Pullicino VN, Lalam RK. Magnetic reso- O, et al. Comparison of irradiation for tomosynthesis and CT
nance imaging of spinal infection. Top Magn Reson Imaging of the wrist. J Radiol 2011;92:32—9.
2007;18:213—22. [66] Gervaise A, Esperabe-Vignau F, Naulet P, Pernin M, Portron
[47] Memarsadeghi M, Breitenseher MJ, Schaefer-Prokop C, Weber Y, Lapierre-Combes M. Évaluation des connaissances des pre-
M, Aldrian S, Gäbler C, et al. Occult scaphoid fractures: CT scripteurs de scanner en matière de radioprotection des
versus MR imaging. Radiology 2006;240:169—76. patients. J Radiol 2011;92:681—7.
[48] Liu PT, Chivers FS, Roberts CC, Schultz CJ, Beauchamp [67] Subhas N, Freire M, Primak AN, Polster JM, Recht MP, Davros
CP. Imaging of osteoid osteoma with dynamic gadolinium- WJ, et al. CT arthrography: in vitro evaluation of single and
enhanced MR imaging. Radiology 2003;227:691—700. dual energy for optimization of technique. Skeletal Radiol
[49] Argin M, Isayev H, Kececi B, Arkun R, Sabah D. Multidetector 2010;39:1025—31.
row computed tomographic angiography findings of muscu- [68] Nakaura T, Awai K, Oda S, Yanaga Y, Namimoto T, Harada
loskeletal tumors: retrospective analysis and correlation with K, et al. A low-kilovolt (peak) high tube current technique
surgical findings. Acta Radiol 2009;50:1150—9. improves venous enhancement and reduces the radiation dose
[50] Thévenin FS, Drapé JL, Biau D, Campagna R, Richarme D, at indirect multidetector row CT venography: initial experi-
Guerini H, et al. Assessment of vascular invasion by bone and ence. J Comput Assist tomogr 2011;35:141—7.
soft tissue tumors of the limbs: usefulness of MDCT angiogra- [69] Mulkens TH, Marchal P, Daineffe S, Salgado R, Bellinck P, te
phy. Eur Radiol 2010;20:1524—31. Rijdt B, et al. Comparison of low-dose with standard-dose
[51] Omoumi P, Mercier GA, Lecouvet F, Simoni P, Vande Berg multidetector CT in cervical spine trauma. Am J Neuroradiol
BC. CT arthrography, MR arthrography, PET, and scintig- 2007;28:1444—50.
raphy in osteoarthritis. Radiol Clin North Am 2009;47: [70] Kröpil P, Fenk R, Fritz LB, Blondin D, Kobbe G, Mödder U,
595—615. et al. Comparison of whole-body 64-slice multidetector com-
[52] Wyler A, Bousson V, Bergot C, Polivka M, Leveque E, Vicaut E, puted tomography and conventional radiography in staging of
et al. Comparison of MR arthrography and CT arthrography multiple myeloma. Eur Radiol 2008;18:51—8.
in hyaline cartilage-thickness measurement in radiographi- [71] Abul-Kasim K, Gunnarsson M, Maly P, Ohlin A, Sundgren PC.
cally normal cadaver hips with anatomy as gold standard. Radiation dose optimization in CT planning of corrective scoli-
Osteoarthritis Cartilage 2009;17:19—25. osis surgery. A phantom study. Neuroradiol J 2008;21:374—82.
[53] Moser T, Dosch JC, Moussaoui A, Dietemann JL. Wrist liga- [72] Damilakis J, Adams JE, guglielmi G, Link TM. Radiation expo-
ment tears: evaluation of MRI and combined MDCT and MR sure in X-ray-based imaging techniques used in osteoporosis.
arthrography. AJR Am J Roentgenol 2007;188:1278—86. Eur Radiol 2010;20:2707—14.
[54] Kim YJ, Choi JA, Oh JH, Hwang SI, Hong SH, Kang HS. Supe- [73] Bohy P, de Maertelaer V, Roquigny A, Keyzer C, Tack D,
rior labral anteroposterior tears: accuracy and interobserver Genevois PA, et al. in patients suspected of having lumbar
388 A. Gervaise et al.

disk herniation: comparison of standard-dose and simulated [92] Moojen TM, Snel JG, Ritt MJ, Venema HW, Kauer JM, Bos KE.
low-dose techniques. Radiology 2005;244:524—31. In vivo analysis of carpal kinematics and comparative review
[74] McCollough CH, Bruesewitz MR, Kofler JM. CT dose reduction of the literature. J Hand Surg Am 2003;28:81—7.
and dose management tools: overview of available options. [93] Foumani M, Strackee SD, Jonges R, Blankevoort L, Zwin-
Radiographics 2006;26:503—12. derman AH, Carelsen B, et al. In vivo three-dimensional
[75] van Straten M, Deak P, Shrimpton PC, Kalender WA. The effect carpal bone kinematics during flexion-extension and radio-
of angular and longitudinal tube current modulations on the ulnar deviation of the wrist: dynamic motion versus step-wise
estimation of organ and effective doses in X-ray computed static wrist positions. J Biomech 2009;42:2664—71.
tomography. Med Phys 2009;36:4881—9. [94] Berdia S, Short WH, Wermer FW, Green JK, Panjabi M.
[76] Mulkens TH, Bellinck P, Baeyaert M, Ghysen D, Van The hysteresis effect in carpal kinematics. J Hand Surg Am
Dijck X, Mussen E, et al. Use of an automatic exposure 2006;31:594—600.
control mechanism for dose optimization in multidetec- [95] Short WH, Werner FW, Fortino MD, Mann KA. Analysis of the
tor row CT examinations: clinical evaluation. Radiology kinematics of the scaphoid and lunate in the intact wrist joint.
2005;237:213—23. Hand Clin 1997;13:93—108.
[77] Mastora I, Rémy-Jardin M, Seuss C, Scherf C, Guillot JP, Rémy [96] Hristova L, Batch T, Blum A. Analyse des artefacts de mouve-
J. Dose reduction in spiral CT angiography of thoracic outlet ment de l’exploration dynamique et volumique au scanner
syndrome by anatomically adapted tube current modulation. 320 barrettes (abstract). J Radiol 2009;90:1583.
Eur Radiol 2001;11:590—6. [97] Blum A, Lecocq S, Roch D, Louis M, Batch T, Dab F, et al.
[78] Nagel HD. CT parameters that influence the radiation dose. Étude cinématique du poignet en 3D et 4D avec un scanner
In: Tack D, Genevois PA, editors. Radiation dose from adult 320 canaux. In: SIMS Opus XXXVI Poignet et main. Montpellier:
and pediatric multidetector computed tomography. Berlin: Sauramps médical; 2009 [p. 375—89].
Springer; 2007. p. 51—79. [98] Tay SC, Primak AN, Fletcher JG, Schmidt B, Amrami KK, Berger
[79] Stradiotti P, Curti A, Castellazzi G, Zerbi A. Metal-related RA, et al. Four-dimensional computed tomographic imaging in
artifacts in instrumented spine. Techniques for reducing arti- the wrist: proof of feasibility in a cadaveric model. Skeletal
facts in CT and MRI: state of the art. Eur Spine J 2009;18: Radiol 2007;36:1163—9.
S102—8. [99] Snel JG, Venema HW, Moojen TM, Ritt JP, Grimbergen CA, den
[80] von Falck C, Galanski M, Shin H. Sliding-thin-slab aver- Heeten GJ. Quantitative in vivo analysis of the kinematics
aging for improved depiction of low-contrast lesions with of carpal bones from three-dimensional CT images using a
radiation dose savings at thin-section CT. Radiographics deformable surface model and a three-dimensional matching
2010;30:317—26. technique. Med Phys 2000;27:2037—47.
[81] McNitt-Gray MF. AAPM/RSNA physics tutorial for resi- [100] Gurung J, Khan MF, Maataoui A, Herzog C, Bux R, Bratzke H,
dents: topics in CT. Radiation dose in CT. Radiographics et al. Multislice CT of the pelvis: dose reduction with regard to
2002;22:1541—53. image quality using 16-row CT. Eur Radiol 2005;15:1898—905.
[82] Walter F, Ludig T, Iochum S, Blum A. Multidectector CT in [101] Levine E, Neff JR. Dynamic computed tomography scanning
musculoskeletal disorders. JBR-BTR 2003;86:6—11. of benign bone lesions: preliminary results. Skeletal Radiol
[83] Blum A. Scanner volumique multicoupe : principes applica- 1983;9:238—45.
tions et perspectives. JBR-BTR 2002;85:82—99. [102] Miles KA, Charnsangavej C, Lee F, Fishman E, Horton K, Lee
[84] Hara AK, Paden RG, Silva AC, Kujak JL, Lawder HJ, Pavlicek TY. Application of CT in the investigation of angiogenesis in
W. Iterative reconstruction technique for reducing body radi- oncology. Acad Radiol 2001;7:840—50.
ation dose at CT: feasibility study. AJR Am J Roentgenol [103] Ketelsen D, Horger M, Buchgeister M, Fenchel M, Thomas C,
2009;193:764—71. Boehringer N, et al. Estimation of radiation exposure of 128-
[85] Silva A, Lawder H, Hara A, Kujak J, Pavlicek W. Innovations in slice 4D-perfusion CT for the assessment of tumor vascularity.
CT dose reduction strategy: application of the adaptive statis- Korean J Radiol 2010;11:547—52.
tical iterative reconstruction algorithm. AJR Am J Roentgenol [104] Heck O, Louis M, Wassel J, Lecocq S, Gondim-Teixeira P, Moisei
2009;194:191—9. A, et al. Apport du scanner volumique dynamique dans le diag-
[86] Gervaise A. Impact des algorithmes itératifs de reconstruction nostic d’ostéome ostéoïde. Paris: Poster journées françaises
sur l’interprétation des images en scanographie. 50e Journées de radiologie; 2010.
scientifiques de la SFPM. Nantes, 9 juin 2011. [105] Lefevre JE, Harbourg B. Mieux voir, plus rapidement et avec
[87] Boas FE, Fleischmann D. Evaluation of two iterative tech- moins de dose. Rapport de la Société française de radiologie
niques for reducing metal artifacts in computed tomography. 2010; [Link]
Radiology 2011;259:894—902. [106] Flohr TG, McCollough CH, Bruder H, Petersilka M, Gruber K,
[88] Buckwalter KA, Lin C, Ford JM. Managing postopera- Süss C, et al. First performance evaluation of a dual-source
tive artifacts on computed tomography and magnetic CT (DSCT) system. Eur Radiol 2006;16:256—68.
resonance imaging. Semin Musculoskelet Radiol 2011;15: [107] Choi HK, Al-Arfaj AM, Eftekhari A, Munk PL, Shojania K, Reid
309—19. G, et al. Dual energy computed tomography in tophaceous
[89] Stierstorfer K, Kuhn U, Wolf H, Petersilka M, Suess C, Flohr gout. Ann Rheum Dis 2009;68:1609—12.
T. Principle and performance of a dynamic collimation tech- [108] Nicolaou S, Yong-Hing CJ, Galea-Soler S, Hou DJ, Louis L,
nique for spiral CT. (abstract) In: Radiological Society of North Munk P. Dual energy CT as a potential new diagnostic tool
America scientific assembly and annual meeting program. Oak in the management of gout in the acute setting. AJR Am J
Brook, IL: Radiological Society of North America, 2007, SSA16- Roentgenol 2010;194:1072—8.
04. [109] Pache G, Krauss B, Strohm P, Saueressig U, Blanke P, Bulla S,
[90] Christner JA, Zavaletta VA, Eusemann CD, Walz-FlanniganAI, et al. Dual energy CT virtual non-calcium technique: detec-
McCollough CH. Dose reduction in helical CT: dynami- ting post-traumatic bone marrow lesions-feasibility study.
cally adjustable z-axis X-ray beam collimation. AJR Am J Radiology 2010;256:617—24.
Roentgenol 2010;194:W49—55. [110] Bamberg F, Dierks A, Nikolaou K, Reiser MF, Becker CR, John-
[91] Wolfe SW, Neu C, Crisco JJ. In vivo scaphoid, lunate and cap- son TRC. Metal artefact reduction by dual energy computed
itate kinematics in flexion and in extension. J Hand Surg Am tomography using monoenergetic extrapolation. Eur Radiol
2000;25:860—9. 2011;21:1424—9.
 Chapitre 4

Article 3 : Scanner ostéo-articulaire à large système de détection : principes, techniques et

applications en pratique clinique et en recherche.

Teixeira P, Gervaise A, Louis M, Lecocq S, Raymond A, Aptel S, Blum A. Musculoskeletal Wide detector CT :
Principles, Techniques and Applications in Clinical Practice and Research. Eur J Radiol 2015; 84:892-900.

171
172
 European Journal of Radiology 84 (2015) 892–900

Contents lists available at ScienceDirect

European Journal of Radiology

journal homepage: [Link]/locate/ejrad

Review

Musculoskeletal wide detector CT: Principles, techniques and

applications in clinical practice and research
Pedro Augusto Gondim Teixeira a,∗ , Alban Gervaise b , Matthias Louis a , Sophie Lecocq a ,
Ariane Raymond a , Sabine Aptel a , Alain Blum a
a
Guilloz Imaging Department, Central Hospital, University Hospital Center of Nancy, 29 avenue du Maréchal de Lattre de Tassigny, 54035 Nancy Cedex,
France
b
Medical Imaging Department, Legouest Military Instruction Hospital, 27 Avenue de Plantières, BP 90001, 57077 Metz Cedex 3, France

a r t i c l e i n f o a b s t r a c t

Article history: A progressive increase in the detector width in CT scanners has meant that advanced techniques such as
Received 28 August 2014 dynamic, perfusion and dual-energy CT are now at the radiologist’s disposal. Although these techniques
Received in revised form may be important for the diagnosis of various musculoskeletal diseases, data acquisition and interpreta-
15 December 2014
tion can be challenging. This article offers a practical guide for the use of these tools including acquisition
Accepted 31 December 2014
protocol, post-processing options and data interpretation based on 7 years of clinical experience in a
tertiary university hospital.
Keywords:
© 2015 Elsevier Ireland Ltd. All rights reserved.
Wide area detector CT
Musculoskeletal
CT perfusion
Dynamic CT
Bone subtraction
Dualenergy

1. Introduction area-detector CT (WADCT), herein defined as multi detector CT

scanners with more than 80 mm of detector width, allows more
In the past, the increased availability of MRI and growing frequent use of sequential acquisition, which has three major
awareness of dose-related patient hazards limited the use of advantages with respect to helical acquisition: higher temporal
musculoskeletal CT in clinical practice. However, more recently resolution, dose-reduction (e.g. no over-ranging and a reduction
multiple advances in CT technology resulted in a “comeback”, sup- of the over-beaming contribution to total exposure), and easier
ported by a considerable decline in radiation exposure and the data post-processing [7,8]. The latter relates to the absence of slice
establishment of clear guidelines and dose limits for clinical use interpolation and to the fact that every voxel of a given volume is
[1–4]. Multi detector CT scanners are now widely available. Mus- acquired at almost the same time (temporal uniformity).
culoskeletal (MSK) CT studies are frequently performed, last only a Dose saving technology implemented by all manufacturers
few seconds and are little more irradiating than conventional radio- (both software and hardware) has led to a two to four fold decrease
graphs. Furthermore, CT has a better diagnostic performance than in the dose delivered in a CT examination and is in great part
radiography for the evaluation of patients with a prosthesis and responsible for the increase in clinical availability of advanced CT
the number of CT studies performed in this setting is increasing applications [2]. Iterative reconstruction (IR) is probably the most
steadily [5]. significant of these developments. IR algorithms are now standard
In the middle of past decade the increase in the number of for most commercial CT scanners and can reduce roughly by half the
detector rows led to an increase in detector width (slice wars) dose maintaining the same image quality [7,9–11]. IR is most effec-
[6]. Coverage has increased from 2 mm with 4 detector-rows up tive when acquisitions have low contrast to noise ratio, which is
to 160 mm with 320 detector-rows. This tendency is apparent particularly interesting for low dose multiphasic protocols [11,12].
in machines from all the major CT manufacturers, with 3 out of WADCT is suited to dynamic study of joints, which can
4 selling scanners with a Z-axis coverage over 80 mm [6]. Wide be imaged in wide ranges of motion using sequential acquisi-
tion. Dynamic CT is the only clinically available technique that
∗ Corresponding author. Tel.: +33 3 83 85 21 61; fax: +33 3 83 85 97 25. allows volumetric study of bone and intra-articular ligaments
E-mail address: ped gt@[Link] (P.A. Gondim Teixeira). during physiologic motion or under stress. This helps overcome

[Link]
0720-048X/© 2015 Elsevier Ireland Ltd. All rights reserved.
 P.A. Gondim Teixeira et al. / European Journal of Radiology 84 (2015) 892–900 893

some limitations of evaluation of dynamic pathology with static low effective doses (0.04 ± 0.05 and 0.05 ± 0.06 mSv for intermit-
imaging and fluoroscopy. Moreover, WADCT is likely to increase tent and continuous acquisition respectively). The effective dose is
the diagnostic performance of CT perfusion, by the use of associated much higher for proximal joints such as hip and shoulder (mean
techniques such as digital subtraction angiography (DSA)-like bone 5.5 ± 1.8 and 6.2 ± 1.2 mSv respectively) where dynamic CT should
subtraction. WADCT also increases the availability of dual-energy, be performed only in selected patients.
which can be performed without hardware features dedicated for Dynamic CT can be coupled with arthrography for the analysis
dual-energy acquisitions (sandwich detectors, dual source, rapid of intra-articular ligaments.
kV switching). To be reliable and reproducible, dynamic CT requires proper
Although the primary role of MRI in identifying and char- patient training and direct visual surveillance of the acquisition.
acterizing MSK pathologic processes remains undisputed, new Prior to the examination, patients are trained to focus on quality of
applications of WADCT may provide additional information and motion and timing. Stabilization of adjacent body parts is impor-
increase the diagnostic performance of CT. Against that back- tant to allow a single movement in one plane of the space to be
ground, increased knowledge of the indications of advanced performed. Avoiding unwanted motion is critical to the interpreta-
WADCT techniques and of data acquisition and post-processing is tion and post-processing of dynamic data. For optimal quality, the
of relevance. This article offers practical guidelines for the applica- motion has to be controlled and smooth, with each maneuver last-
tion of dynamic CT, CT perfusion and dual-energy for the evaluation ing between 10 and 15 s. Patient guidance and surveillance systems
of MSK pathology. are now available to optimize the quality of dynamic acquisitions
[17].
2. Materials and methods The speed of motion that can be evaluated with dynamic CT
is limited. Faster movements can lead to significant degradation
From March 2008 to December 2013 a 320 detector-row CT of image quality [18]. Excessive motion artifacts appear as ghost-
scanner (Aquilion ONE, Toshiba Medical Systems, Otawara, Japan) ing and linear streaks and are frequently located over 5 cm of the
has been used for the evaluation of patients with various types fulcrum of motion (where linear speeds are higher), with limited
of musculoskeletal disorders (traumatic, degenerative, inflamma- impact on the interpretation of more proximal structures. There
tory and tumoral) at a tertiary university hospital by a team of are two ways of reducing motion artifacts: slowing patient motion
musculoskeletal radiologists (seven physicians). An average of 30 or increasing volume acquisition speed. Tube rotation speed, has
advanced CT studies (Dynamic CT, perfusion CT and dual-energy a strong impact on image quality and should be kept as high as
CT) are performed weekly in our institution. The team’s experience possible to avoid motion artifacts. Partial scanning and high pitch
with the used of advanced CT techniques is reported along with a techniques (e.g. dual source CT) can provide full volume reconstruc-
systematic literature review. tion with less than 360◦ of tube rotation with can greatly reduce
volume acquisition time [19].
3. Dynamic CT Dynamic CT offers considerable advantage over imaging the
extremes of joint range, since dynamic abnormalities can be
Dynamic CT uses multiple, low-dose, sequential acquisitions of limited to a particular maneuver or motion path. Punctual motion
the same anatomic area during motion. This technique requires anomalies can be identified with dynamic CT. In cases of patellar
wide detector systems to be performed for it strongly relies on high maltracking, abnormal motion is present at some point of the first
temporal resolution, which is possible with wide detector CT using 30◦ of knee flexion when the patella engages the trochlea. Similarly,
sequential acquisition (all portions of the volume must be imaged in cases of midcarpal instability there is abrupt, abnormal motion
at the same time). A single movement or stress maneuver is studied of the whole first carpal row at some point of radio-ulnar deviation
per acquisition. Although this technique is most frequently used for or dart throwing maneuver [20]. In the wrist, there are variations
the evaluation of the wrist, it can be used on various joints (shoul- in bone position (or angulation) that are specific of a motion path, a
der, hip, elbow, knee, and ankle) [13]. Multiple maneuvers have phenomenon called hysteresis. The hysteresis of a bone (e.g. lunate)
been described in the literature, especially for the wrist [14–16]. is increased in cases of wrist instability [21].
With respect to fluoroscopy, dynamic CT allows multiplanar and Analysis of dynamic CT data can be challenging. Image inter-
3-D study of bone and intra-articular ligaments overcoming con- pretation on a workstation that allows multiplanar and volume
trast and superimposition issues with the former technique at the rendering (VR) of multiple volumes is recommended. Soft tis-
expense of a lower temporal resolution. sue kernels yield better results for VR reconstruction while sharp
Two scanning modes can be used for the acquisition of dynamic bone kernels are recommended for multiplanar reconstructions.
data: VR images are used for the initial study of inter-bone relations. The
Intermittent acquisition (volumes separated by a variable time findings have to be confirmed on multiplanar reformats, which are
interval) is used by default as it limits radiation exposure. The inter- also used for the study of intra-articular ligaments with dynamic CT
volume interval is usually 1 s and 10–15 volumes are sufficient to arthrography. Dedicated analysis software allowing bone locking
image most of the dynamic pathology encountered. This technique (e.g. after volume registration, all motion is displayed with respect
is not suitable for the evaluation of fast or jerky movements espe- to a single bone, which remains static or locked) and the automatic
cially those whose speed cannot be controlled by the patient for it propagation of distance and angular measurement throughout the
lacks sufficient temporal resolution. study volumes is currently undergoing clinical testing.
Continuous acquisition (no interval between volumes) offers Dynamic CT has three main applications:
maximal temporal resolution at the expense of a higher dose. Tem-
poral resolutions of 200 ms or lower can be achieved with this 1. Entrapment, impingement and snapping. Because of the multi-
method by using partial volume reconstruction techniques. Contin- ple possible etiologies of these syndromes, it may be difficult
uous acquisitions should not last more than 5 s in order to maintain to ascertain clinically which structure is responsible for patient
dose exposure within acceptable limits. This acquisition mode is symptoms. With static imaging, the diagnosis of snapping and
recommended in cases of joint or ligamentous snapping, in which, impingement syndromes is based solely on secondary findings,
the symptoms cannot be reproduced with a slow controlled motion. which can be insufficient. The advantage of dynamic studies is
Regardless of the acquisition method dynamic CT can be per- that the zone of impingement is directly demonstrated, help-
formed in the extremities (ankle, knee, wrist, elbow), with very ing confirm the diagnosis and guide surgical therapy (Fig. 1).
894 P.A. Gondim Teixeira et al. / European Journal of Radiology 84 (2015) 892–900

Fig. 1. 39 year-old female with right scapular pain and grinding. No anomalies were identified on conventional imaging studies. Images 1–3 were extracted from a dynamic
CT study using continuous acquisition. The superior scapular angle impinges against the second rib at the end of shoulder girdle rotation (red circle). The acquisition lasted
5 s and the temporal resolution after post-processing was 200 ms. Supplementary material – video 1.

In this context, dynamic CT is indicated when the diagnosis is

not reached with conventional methods, particularly in the pre-
operative evaluation.
Dynamic CT angiography can be performed for the evaluation
of neuro-vascular entrapment syndromes, such as thoracic out-
let syndrome or popliteal artery entrapment, to detect bony or
soft tissue compression. With WADCT, multiple volumes can be
acquired during the first pass of a single contrast bolus, which
allows multiple maneuvers (e.g. gastrocnemius contraction and
hamstring contraction for popliteal artery entrapment) to be
performed with a single injection. Additionally, CT angiogra-
phy can be performed during head rotation or other maneuvers
eliciting patient symptoms, for the evaluation of thoracic out-
let syndrome, increasing the sensitivity for the detection of
impingement. The amount of contrast injected is reduced when
compared to conventional CT angiography, which requires one
contrast bolus per maneuver or position evaluated (Fig. 2).
2. Evaluation of the sufficiency of intra-articular ligaments. A lig-
ament is considered sufficient when it is able to withstand
physiological stress without significant elongation. Ligament
sufficiency cannot be evaluated with static imaging [22]. With
dynamic CT arthrography, some ligaments (e.g. scapho-lunate
and luno-triquetral ligaments) can be evaluated under stress.
The main application of this technique is the evaluation of
dynamic carpal instability (Fig. 3).
3. Analysis of complex motion. In some joints, such as the wrist and
the subtalar joint, motion is too complex (e.g. multiple moving
structures, rotatory motion) and even static evaluation at the
extremes of joint amplitude (regardless of the imaging method:
radiographs, CT or MRI) is not sufficient to demonstrate dynamic
pathologic processes [23]. Dynamic CT is able to identify brief
changes in bone relation during the course of a given motion,
making it particularly interesting for the evaluation of complex
movements.

Table 1 demonstrates the acquisition protocol for the most fre-

quently used maneuvers for various joints and their indications.
Fig. 2. 35 year-old female with a chronic bilateral paresthesia of the superior limbs.
Dynamic CT angiography was performed with the arms above the shoulder line
4. CT perfusion during left to right head motion. Volume rendered upper view of the thoracic outlet
with the head turned to the left, front and right (A, B and C respectively). Note the
The aim of CT perfusion, also based on low-dose intermittent stenosis of the subclavian artery on the extremes of head motion (arrows in A and
acquisition of a target area, is analysis of the first passage of the C), confirming the diagnosis of thoracic outlet syndrome. No compression is seen in
neutral position (B).
contrast medium bolus. The use of CT perfusion can be facilitated
by a wide detector system, but this technique can be performed
with most CT scanner models. Perfusion can provide useful hemo- testing, the effectiveness of CT guided ablation could be confirmed
dynamic information in cases of vascular tumor or micro-traumatic with intraoperative CT perfusion, which demonstrated the absence
vascular injury (e.g. hypothenar hammer syndrome) (Fig. 4). Similar of nidus enhancement immediately after ablation (Fig. 5).
to MRI perfusion, CT perfusion can be useful for tumor charac- It is important to keep in mind that MRI remains the method of
terization, differentiation between residual/recurrent tumor and choice for perfusion studies as it has a higher sensitivity for contrast
post-operative fibrosis and biopsy targeting [24,25]. In preliminary detection and uses no ionizing radiation. CT perfusion is, however,
 P.A. Gondim Teixeira et al. / European Journal of Radiology 84 (2015) 892–900 895

Fig. 3. 34 year-old male with scaphoid fracture and rupture of the scapholunate ligament treated surgically with persistent wrist pain. Figs. 1–3 are a series of three coronal
images extracted from a dynamic CT arthrography. In the neutral position (Image 1) there is a normal scapholunate distance and a thickened but continuous scapholunate
ligament (arrow). Throughout ulnar deviation (Images 2 and 3) there is thinning and elongation of the scapholunate ligament, that is insufficient, allowing scapholunate
diastasis (arrows). Supplementary material – video 2.

Table 1
Protocol recommendations and indications for dynamic CT.

Joint Maneuver Indication Study type Z-axis coverage (mm) Tube outputa Scanning mode

Radio-ulnar Ulno-lunar Dynamic CT 80 80 kVp, 100 mAs Intermittent

deviation impingement,
Wrist
carpal instability
Clenching fist Dissociative carpal Dynamic CT arthrography 60 80 kVp, 100 mAs Continuous
examination instability
Dart throwing Carpal instability Dynamic CT 80 80 kVp, 100 mAs Intermittent
Prono-supination DRUJb instability Dynamic CT 60 80 kVp, 100 mAs Intermittent

Scapulo-thoracic Shoulder blade Snapping scapula Dynamic CT 140 or higher 100 kVp, 200 mAs Continuous
rotation (reproduce syndrome
patient symptoms)

Thoracic outlet Head rotation with Thoracic outlet Dynamic CT angiography 80–100 100 kVp, 150 mAs Intermittent
arms over syndrome
shoulders

Flexion-internal Femoro-acetabular Dynamic CT 80 100 kVp, 200 mAs Intermittent

Hip
rotation impingement
Flexion-extension Lateral snapping Dynamic CT 80 100 kVp, 200 mAs Intermittent
hip

Flexion-extension Patellar Dynamic CT 140 or higher 100 kVp, 150 mAs Intermittent
Knee
(comparative) maltracking
Sural triceps and Popliteal Dynamic CT angiography 140 or higher 100 kVp, 150 mAs Intermittent
hamstrings entrapment
contraction syndrome
(sequentially)
a
Susceptible to changes according to patient body habitus.
b
DRUJ = distal radio-ulnar joint.

a valuable option when MRI is contra-indicated or unavailable. Semi-quantitative analysis is based on the construction of a
Despite these advantages of MRI over CT, in clinical practice, CT time-to-density graph, which yields various perfusion parameters
and MRI perfusion have a similar diagnostic performance [26,27]. useful for lesion characterization as demonstrated in Fig. 7 [25].
The main advantage of CT over MRI relates to the linear asso- Although semi-quantitative parameters are easy to obtain, intra-
ciation between iodine concentration and CT number, which is and inter-patient comparisons should be performed with caution
not the case for MRI and gadolinium. This feature makes density- even with similar injection protocols. Some time-related param-
related perfusion parameters easier to calculate and compare eters (e.g. time-to-peak, delay of enhancement) are dependent on
[28]. the anatomic location of the tumor. Density-related parameters are
Three types of perfusion parameters are available for analysis: dependent on tube output (kVp), scanner calibration, iodine con-
visual, semi-quantitative and quantitative. centration of the contrast media and injection rate. One of the most
Visual analysis is performed by direct comparison of tumor reliable semi-quantitative parameters is the delay between arterial
enhancement with nearby arterial enhancement on 4-D series dis- and tumor peaks, which can be compared directly in most patients.
played on soft tissue windowing. Tumor enhancement should be Quantitative analysis is based on estimation of the tissue con-
evaluated for density and timing using arterial enhancement as a centration of iodine using mathematical pharmacokinetic models –
reference. Additionally, the presence of intra-tumoral neovessels most frequently bicompartmental single arterial input models such
can be evaluated with either conventional or VR reconstructions. as Brix’s or Toft’s models [30,31]. Parameters related to capillary
Neovessels are intimately associated with tumor aggressiveness density, and vessel permeability can be calculated. The most fre-
and are characterized by an erratic variation in caliber that leads to quently calculated quantitative parameters are plasmatic volume
the formation of a non-organized intra-tumoral vascular network (Vp %), volume transfer constant form the plasma to the extravas-
(Fig. 6) [29]. cular extracellular space (EES) (Ktrans ), rate constant of the backflux
896 P.A. Gondim Teixeira et al. / European Journal of Radiology 84 (2015) 892–900

Fig. 4. 58 year-old male smoker with persistent coldness and discoloration of the 4th and 5th fingers. (A–C) CT perfusion sequential MIP images of the hand and wrist. There
is a late filling of the ulnar artery (arrowheads) with an obstruction at the level of the pisiform bone (fat arrows). Collateral vessels at the level of Guyon’s canal are seen in the
latter phases (thin arrows). The proper digital arteries of the fifth finger are not seen (squiggly arrows). Note that the superficial and deep palmar arches are filled through
the radial artery.

from the EES to the plasma (kep ) and EES volume (Ve %). Compared Table 2
CT perfusion acquisition protocol recommendations.
to brain, quantitative perfusion parameters in body CT perfusion
require more complex (and hence less precise) calculation methods CT perfusion MSK
because outside the blood–brain barrier normal capillaries allow Tube output Adapted to anatomy and patient
the passage of contrast medium molecules. Quantitative parame- body habitus
ters are highly dependent on the model used for calculation, but Z-axis coverage 40–160 mma
inter-patient comparisons are possible provided the model and Slice thickness 0.5 mm
Injection rate 5 ml/s
acquisition parameters are the same. Quantitative perfusion is a
Contrast volume 2 ml/kgb
subject of intense research. It requires dedicated software and fur- Bolus tracking Yes
ther studies are still necessary to establish diagnostic criteria for Number of phases 18
MSK pathology. Inter volume delay – arterial phase 5s
Inter volume delay – venous phase 10 s
There is no consensus on the best acquisition protocol
a
[32,33]. Table 2 demonstrates the parameters of the CT perfusion Z-axis coverage should be kept as low as possible to avoid unnecessary radiation
protocol used in our institution. Timing is crucial for perfusion and exposure.
b
Up to a maximum of 150 ml.
the use of bolus tracking is recommended. In our experience, with
WADCT and using an adequate protocol, CT perfusion can usually
be performed with a DLP under 800 mGy cm. This is a lower level
of exposure than abdominal or thoracic MDCT with helical acqui- very difficult to identify visually. Classic bone subtraction does not
sition. CT perfusion is particularly interesting in the extremities, allow evaluation of intra-osseous enhancement. Digital subtraction
where it can be performed with very low effective doses – usually angiography (DSA)-like bone subtraction uses a pre-contrast mask
under 0.5 mSv. In the trunk and pelvic and shoulder girdles effective volume that can be registered and subtracted from one or more
dose levels are still prohibitive [34]. post-contrast volumes provided the same acquisition parameters
are used (Fig. 8). With DSA-like bone subtraction enhancement of
densely calcified or non-lytic bone lesions can be seen on CT.
4.1. Digital subtraction angiography-like bone subtraction DSA-like bone subtraction offers a significant improvement
in the visualization of intra-osseous enhancement. Teixeira et al.
Calcification, bone and iodinated contrast medium share reported a sensitivity of over 70% and a specificity of 100% for the
a similar density in Hounsfield units. Although intra-osseous identification of bone marrow edema pattern (BMEP) adjacent to
enhancement is measurable with conventional CT images, it is bone tumors using MRI as standard of reference [35]. DSA-like bone
 P.A. Gondim Teixeira et al. / European Journal of Radiology 84 (2015) 892–900 897

Fig. 5. 16 year-old male with a trabecular osteoid osteoma of the medial femoral condyle of the left knee. (A) Intraoperative axial oblique CT perfusion image performed
immediately after percutaneous laser therapy (500 J for 3 min). The needle track (arrows) is seen confirming the adequate needle position with respect to the nidus (arrowhead).
Saline–iodine solution was used as a coolant agent for chondral protection (*). (B) Time-to-density graph obtained after laser therapy showing no enhancement in the nidus
(purple curve) and popliteal artery enhancement (green curve). (C) CT perfusion time-to-density graph obtained 1 week before treatment showing typical steep and early
nidus enhancement (purple curve).

Fig. 7. CT perfusion of a grade III soft tissue sarcoma of the right arm of a 65 year-
old male. Time-to-density graph comparing the tumor enhancement (green curve)
with that of the brachial artery (purple curve). Tumor and artery enhancement peaks
are indicated by the fat and thin arrows, respectively. The thin white lines indicate
the most frequently used time-related semi-quantitative parameters (time-to-peak
Fig. 6. 51 year-old male with a biopsy-proven fibrosarcoma. 3-D volume rendered
and tumor-arterial peak delay). The enhancement slope is demonstrated by the red
image at the late arterial phase demonstrating the feeding vessels at the superior
line over the tumor enhancement curve. Finally, the area under the tumor curve
and inferior poles of a soft tissue mass (arrowheads). Neovessels are seen inside
represents the enhancement integral and is related to tumor capillary density.
the mass (arrows). Note the tortuosity and the erratic variation in caliber of the
intra-tumor vessels.
898 P.A. Gondim Teixeira et al. / European Journal of Radiology 84 (2015) 892–900

Fig. 8. Schematic representation of the DSA-like bone subtraction procedure. In the first two steps, the mask volume and the post-contrast volume are acquired. Images (1
and 2) Axial CT of the fifth lumbar vertebra. Note osseous enhancement is very faint and easily missed. Step 3 can be performed any time after acquisition with a dedicated
workstation. Image (4) Axial bone subtracted CT image showing a focal zone of contrast enhancement at the right superior articular facet (arrow).

Fig. 9. 64 year-old female with a sarcomatoid carcinoma of the distal femur, pre-operative evaluation. (A) Sagittal post-contrast CT image viewed in a soft tissue window
setting in the venous phase demonstrating an aggressive lytic bone tumor with irregular margins (arrows), cortical destruction and a soft tissue mass (arrowheads). (B)
Sagittal bone subtracted CT image of the same anatomic region demonstrating enhancement in both the lytic and soft tissue components of this lesion (arrowheads) but also
at the non-lytic bone (white arrows).

Fig. 10. 66 year-old male in chronic dialysis due to a bilateral nephrectomy secondary to a multifocal tubulo-papilary carcinoma. Dual-energy was performed with 80 and
135 kVp. Dual-energy graph and color map for the differentiation between uric acid and calcium carbonate (CaCO3 ). The color map demonstrates the presence of uric acid
in the soft tissue calcification seen at the distal thumb (arrowhead). The green dot in the graph (arrow) depicts the behavior of a region of interest placed on the soft tissue
calcification that is close to that of uric acid (red line).
 P.A. Gondim Teixeira et al. / European Journal of Radiology 84 (2015) 892–900 899

subtraction is particularly useful for the evaluation of small lytic [5] Miller TT. Imaging of hip arthroplasty. Eur J Radiol 2012;81(December
bone lesions and for the diagnosis of osteomyelitis (Fig. 9). (12)):3802–12.
[6] Parrish FJ. Volume CT: state-of-the-art reporting. Am J Roentgenol
2007;189(September (3)):528–34.
5. Dual-energy [7] Gervaise A, Osemont B, Lecocq S, Noel A, Micard E, Felblinger J, et al. CT
image quality improvement using adaptive iterative dose reduction with
wide-volume acquisition on 320-detector CT. Eur Radiol 2012;22(February
Dual-energy relies on element-specific attenuation behavior on (2)):295–301.
exposure to X-ray beams with different energies. There are mul- [8] Gervaise A, Teixeira P, Villani N, Lecocq S, Louis M, Blum A. CT dose optimisation
tiple ways of performing dual-energy acquisitions (dual source, and reduction in osteoarticular disease. Diagn Interv Imaging 2013;94(April
(4)):371–88.
sandwich detectors, rapid kVp switching) [36]. With WADCT, dual- [9] Maxfield MW, Schuster KM, McGillicuddy EA, Young CJ, Ghita M, Bokhari SAJ,
energy is performed by the acquisition of two successive volumes of et al. Impact of adaptive statistical iterative reconstruction on radiation dose
the same anatomic region with different tube kVp settings (usually in evaluation of trauma patients. J Trauma Acute Care Surg 2012;73(December
(6)):1404–9.
80 and 135 kVp). Although this technique offers a limited spectral [10] Mitsumori LM, Shuman WP, Busey JM, Kolokythas O, Koprowicz KM. Adaptive
separation and requires specific software, it does not have spe- statistical iterative reconstruction versus filtered back projection in the same
cific hardware requirements. Dual-energy is mainly used in MSK patient: 64 channel liver CT image quality and patient radiation dose. Eur Radiol
2012;22(January (1)):138–43.
for differentiation between intra-articular and peri-articular hyper-
[11] Wang R, Schoepf UJ, Wu R, Reddy RP, Zhang C, Yu W, et al. Image quality and
dense materials, which is useful for the differential diagnosis of radiation dose of low dose coronary CT angiography in obese patients: sinogram
gout. Dual-energy can also be used to differentiate between calci- affirmed iterative reconstruction versus filtered back projection. Eur J Radiol
2012;81(November (11)):3141–5.
fication and iodine on CT arthrography [37]. Additionally, virtual
[12] Desai GS, Uppot RN, Yu EW, Kambadakone AR, Sahani DV. Impact of iterative
non-contrast medium enhanced images can be obtained by selec- reconstruction on image quality and radiation dose in multidetector CT of large
tively subtracting the contrast medium from a contrast-enhanced body size adults. Eur Radiol 2012;22(August (8)):1631–40.
examination [38]. [13] Wassilew GI, Janz V, Heller MO, Tohtz S, Rogalla P, Hein P, et al. Real time
visualization of femoroacetabular impingement and subluxation using 320-
Dual-energy acquisitions are interpreted based on prior slice computed tomography. J Orthop Res 2013;31(February (2)):275–81.
knowledge of the expected behavior of a particular element. Post- [14] Shores JT, Demehri S, Chhabra A. Kinematic “4 dimensional” CT imaging in
processing software can then reconstruct color-coded functional the assessment of wrist biomechanics before and after surgical repair. Eplasty
2013;13:e9.
maps that identify a particular element, most frequently iodine, cal- [15] Halpenny D, Courtney K, Torreggiani WC. Dynamic four-dimensional 320 sec-
cium or uric acid. Data can also be presented in the form of graphs tion CT and carpal bone injury – a description of a novel technique to diagnose
comparing the behavior of an ROI with that of known elements scapholunate instability. Clin Radiol 2012;67(February (2)):185–7.
[16] Edirisinghe Y, Troupis JM, Patel M, Smith J, Crossett M. Dynamic motion analysis
(Fig. 10). of dart throwers motion visualized through computerized tomography and cal-
culation of the axis of rotation. J Hand Surg Eur Vol 2014;39(May (4)):364–72.
6. Conclusion [17] Labone N, Kammacher W, Blum A. Etude préliminaire d’un outil d’aide à
l’acquisition 4D au scanner. Oral presentation. Paris, France: Journées Fran-
caises de Radiologie; 2014.
WADCT offers new diagnostic possibilities for the diagnosis of [18] Fahmi F, Beenen LFM, Streekstra GJ, Janssen NY, de Jong HW, Riordan A, et al.
MSK disorders through the application of dynamic, perfusion and Head movement during CT brain perfusion acquisition of patients with sus-
pected acute ischemic stroke. Eur J Radiol 2013;82(December (12)):2334–41.
dual-energy CT. These techniques are readily available and can be [19] Hou DJ, Tso DK, Davison C, Inacio J, Louis LJ, Nicolaou S, et al. Clinical utility of
performed with various CT scanner models on the market. This arti- ultra high pitch dual source thoracic CT imaging of acute pulmonary embolism
cle offers a practical guide to the clinical application of these tools in the emergency department: are we one step closer towards a non-gated
triple rule out? Eur J Radiol 2013;82(October (10)):1793–8.
and to interpretation of the data they yield, which it is to be hoped
[20] Garcia-Elias M. The non-dissociative clunking wrist: a personal view. J Hand
will increase their use and allow better result comparison in future Surg Eur Vol 2008;33(December (6)):698–711.
research. [21] Berdia S, Short WH, Werner FW, Green JK, Panjabi M. The hysteresis effect in
carpal kinematics. J Hand Surg Am 2006;31(April (4)):594–600.
[22] Van Dyck P, Gielen JL, Vanhoenacker FM, Wouters K, Dossche L, Parizel PM.
Funding Stable or unstable tear of the anterior cruciate ligament of the knee: an MR
diagnosis? Skeletal Radiol 2012;41(March (3)):273–80.
[23] Van de Giessen M, Foumani M, Vos FM, Strackee SD, Maas M, Van Vliet LJ, et al.
This work was not funded.
A 4D statistical model of wrist bone motion patterns. IEEE Trans Med Imaging
2012;31(March (3)):613–25.
Conflict of interest [24] Teixeira PAG, Chanson A, Beaumont M, Lecocq S, Louis M, Marie B, et al. Dynamic
MR imaging of osteoid osteomas: correlation of semiquantitative and quanti-
tative perfusion parameters with patient symptoms and treatment outcome.
Two authors involved in this work, (P.A.G.T. and A.B.) participate Eur Radiol 2013;23(9 (Sep)):2602–11.
on a non-remunerated research contract with Toshiba Medical sys- [25] Gondim Teixeira PA, Lecocq S, Louis M, Aptel S, Raymond A, Sirveaux F, et al.
Wide area detector CT perfusion: can it differentiate osteoid osteomas from
tems for the development and clinical testing of post processing
other lytic bone lesions? Diagn Interv Imaging 2014;95(6 (Jun)).
tools for MSK CT. The other authors have no potential conflicts of [26] Otton J, Morton G, Schuster A, Bigalke B, Marano R, Olivotti L, et al. A direct
interest to disclose. comparison of the sensitivity of CT and MR cardiac perfusion using a myocardial
perfusion phantom. J Cardiovasc Comput Tomogr 2013;7(April (2)):117–24.
[27] De Simone M, Muccio CF, Pagnotta SM, Esposito G, Cianfoni A. Comparison
References between CT and MR in perfusion imaging assessment of high-grade gliomas.
Radiol Med 2013;118(February (1)):140–51.
[1] Lee T-Y, Chhem RK. Impact of new technologies on dose reduction in CT. Eur J [28] Goh V, Padhani AR. Imaging tumor angiogenesis: functional assessment using
Radiol 2010;76(October (1)):28–35. MDCT or MRI? Abdom Imaging 2006;31(April (2)):194–9.
[2] McCollough CH, Chen GH, Kalender W, Leng S, Samei E, Taguchi K, et al. [29] Van Rijswijk CSP, Geirnaerdt MJA, Hogendoorn PCW, Taminiau AHM, van
Achieving routine submillisievert CT scanning: report from the summit on Coevorden F, Zwinderman AH, et al. Soft-tissue tumors: value of static and
management of radiation dose in CT. Radiology 2012;264(August (2)):567–80. dynamic gadopentetate dimeglumine-enhanced MR imaging in prediction of
[3] Berland, Lincoln L, Armstrong, Mark R, Cody Dianna D, Mahadevappa M, malignancy. Radiology 2004;233(November (2)):493–502.
Mayo-Simith William W. ACR practice guideline for performing and interpre- [30] Brix G, Semmler W, Port R, Schad LR, Layer G, Lorenz WJ. Pharmacokinetic
ting diagnostic computed tomography (CT). American College of Radiology; parameters in CNS Gd-DTPA enhanced MR imaging. J Comput Assist Tomogr
2011. Available from: [Link] 1991;15(August (4)):621–8.
guidelines/CT Performing [Link] [Internet]. [31] Tofts PS. Modeling tracer kinetics in dynamic Gd-DTPA MR imaging. J Magn
[4] Bongartz G, Golding SJ, Jurik AG, Leonardi M, van Meerten EvP, Geleijns Reson Imaging 1997;7(February (1)):91–101.
J, et al. European guidelines on quality criteria for computed tomogra- [32] Miles KA. Perfusion CT for the assessment of tumour vascularity: which proto-
phy. European Commission’s Radiation Protection Actions; 1998. Available col? Br J Radiol 2003;76(1):S36–42.
from: [Link] [33] Kambadakone AR, Sharma A, Catalano OA, Hahn PF, Sahani DV. Protocol mod-
Guidelines Quality Criteria Computed Tomography Eur [Link] [Internet]. ifications for CT perfusion (CTp) examinations of abdomen-pelvic tumors:
900 P.A. Gondim Teixeira et al. / European Journal of Radiology 84 (2015) 892–900

impact on radiation dose and data processing time. Eur Radiol 2011;21(June [36] Vrtiska TJ, Takahashi N, Fletcher JG, Hartman RP, Yu L, Kawashima A. Geni-
(6)):1293–300. tourinary applications of dual-energy CT. AJR Am J Roentgenol 2010;194(June
[34] Biswas D, Bible JE, Bohan M, Simpson AK, Whang PG, Grauer JN. Radiation expo- (6)):1434–42.
sure from musculoskeletal computerized tomographic scans. J Bone Joint Surg [37] Glazebrook KN, Guimarães LS, Murthy NS, Black DF, Bongartz T, Manek NJ,
Am 2009;91(August (8)):1882–9. et al. Identification of intraarticular and periarticular uric acid crystals with
[35] Gondim Teixeira PA, Hossu G, Lecocq S, Razeto M, Louis M, Blum A. Bone mar- dual-energy CT: initial evaluation. Radiology 2011;261(November (2)):516–24.
row edema pattern identification in patients with lytic bone lesions using [38] Toepker M, Moritz T, Krauss B, Weber M, Euller G, Mang T, et al. Virtual non-
digital subtraction angiography-like bone subtraction on large-area detector contrast in second-generation, dual-energy computed tomography: reliability
computed tomography. Invest Radiol 2014;49(March (3)):156–64. of attenuation values. Eur J Radiol 2012;81(March (3)):e398–405.
182
Discussion et conclusion du chapitre 4 :

Nos articles montrent comment il est possible d’optimiser et de réduire la dose dans
plusieurs domaines d’applications cliniques et mettent en avant l’intérêt de combiner
l’utilisation des facteurs comportementaux et des facteurs techniques.

Article 1 : Scanner basse dose pour la recherche d’une colique néphrétique : comment
faire en pratique clinique ?

Les scanners réalisés dans le cadre d’une suspicion de colique néphrétique représentent un
cas particulier en imagerie abdominale. Premièrement, ils sont réalisés la plupart du temps
avec une seule acquisition sans injection. Deuxièmement, il est possible de réduire la longueur
d’acquisition en centrant le scanner sur les voies urinaires [46]. Enfin, le bruit de l’image peut
être augmenté de manière significative sans altérer la performance diagnostique. Cela est lié à
l’hyperdensité spontanée de la plupart des calculs urinaires et à l’important RCB entre les
calculs urinaires et les parties molles environnantes [57]. Les doses des scanners réalisés pour
le bilan d’une colique néphrétique peuvent donc être réduites de manière importante, de
l’ordre de 70 à 90 % par rapport à une acquisition abdominopelvienne standard, sans en
altérer la performance diagnostique [60-61]. Cet exemple est intéressant car il met en avant la
relation qu’il peut exister entre la dose, la qualité d’image et la performance diagnostique. Au
final, le but de toute démarche d’optimisation de la dose est de maintenir une excellente
performance diagnostique pour la dose d’irradiation la plus faible possible. Pour cela, la
qualité d’image peut être plus ou moins volontairement dégradée en fonction des structures
qui sont analysées. De plus, cet exemple montre l’intérêt en pratique clinique courante de
combiner l’utilisation des facteurs comportementaux et techniques dans une démarche
d’optimisation de la dose d’irradiation au scanner. Ainsi, tandis que les reconstructions
itératives permettent de réduire environ de moitié la dose à qualité d’image équivalente, la
réduction de la longueur d’acquisition avec un centrage du scanner sur les voies urinaires est
une manière simple pour réduire de façon significative, de l’ordre de 15 à 20 %, la dose
délivrée au patient. Enfin, malgré l’utilisation des algorithmes de reconstructions itératives,
les données de la littérature montrent qu’à ce jour il n’est pas encore possible de réduire la
dose sous le seuil de 1 mSv sans altérer la visualisation des calculs urinaires de moins de 3
mm [62-63]. Bien que n’ayant pas de traduction médicale directe, le franchissement de ce

183
seuil « psychologique » sera peut-être bientôt possible grâce aux nouvelles générations de
reconstructions itératives en cours de développement.

Article 2 : Optimisation et réduction de la dose en scanner ostéo-articulaire.

Le scanner ostéo-articulaire comporte de nombreuses particularités. Tout d’abord, il est

souvent possible de substituer le scanner par un examen d’imagerie non irradiant comme
l’IRM ou l’échographie. Par exemple, à l’HIA Legouest, les scanners lombaires ou cervicaux
des patients de moins de 60 ans adressés pour la recherche d’un conflit disco-radiculaire sont
systématiquement substitués en IRM, en l’absence de contre-indication à celle-ci.
Deuxièmement, le scanner en imagerie ostéo-articulaire regroupe de nombreux types
d’examens : scanner des articulations périphériques ou proximales, scanner du rachis,
arthroscanner, scanner dynamique ou de perfusion, scanner double-énergie. Les doses
délivrées au patient en scanner ostéo-articulaire sont donc très variables, d’un facteur 1 à 1000,
car elles dépendent principalement des longueurs d’acquisition et des coefficients de
conversation tissulaire qui sont très différents d’une localisation anatomique à l’autre [64]. Par
exemple, un scanner d’une articulation périphérique comme le poignet a une dose inférieure à
la dose d’une radiographie thoracique de face (moins de 0,05 mSv) tandis que la dose d’un
scanner lombaire peut aller jusqu’à 5 à 10 mSv. Parallèlement, les modalités d’optimisation et
de réduction de la dose doivent être adaptées en fonction des localisations anatomiques et du
type d’examen réalisé. Ainsi, il faut différencier les scanners dont l’intérêt est principalement
lié à l’analyse de l’os et ceux s’intéressant aux parties molles. Pour les premiers, le bon RCB
permet d’augmenter le bruit de l’image sans altérer la performance diagnostique. Il est donc
possible de réduire de manière importante les doses. Pour les scanners s’intéressant aux
parties molles, comme par exemple les scanners du rachis réalisés pour la recherche d’un
conflit disco-radiculaire, le faible RCB nécessite de maintenir une bonne qualité d’image et de
ce fait les doses délivrées sont plus importantes. Pour ce type de scanners, l’utilisation des
reconstructions itératives et de la modulation automatique du mA sont alors les deux
principales mesures qui permettent d’optimiser la dose.

Article 3 : Scanner ostéo-articulaire à large système de détection : principes, techniques

et applications en pratique clinique et en recherche.

184
 En scanner ostéo-articulaire, de nouvelles applications avancées ont vu le jour ces
dernières années : scanner dynamique 4D des articulations, scanner de perfusion tumorale ou
encore scanner double-énergie. Ces techniques semblent jouer un rôle important dans le
diagnostic de nombreuses pathologies ostéo-articulaires. Toutefois, les scanners dynamiques
et de perfusion nécessitent la répétition de multiples volumes d’acquisition. La maîtrise des
doses délivrées est un facteur primordial pour permettre l’utilisation en pratique clinique de
ces nouvelles applications avancées. Pour les articulations périphériques, les coefficients de
conversion tissulaire pour le calcul des doses efficaces sont très faibles car ces articulations
sont éloignées des organes radio-sensibles. De ce fait, même avec la répétition de 10 à 15
volumes d’acquisition, le centrage de la couverture d’acquisition et la réduction des
paramètres d’acquisition permettent de maintenir des doses très faibles, inférieures à 1 mSv.
A partir de tests que nous avons réalisés sur cadavres, nous pensons qu’il est encore possible
de réduire de manière importante les paramètres d’acquisition et donc la dose des scanners
dynamiques, tout en conservant une qualité d’image suffisante pour l’analyse des
mouvements. Par exemple, pour une acquisition dynamique de la cheville, la dose d’une
phase d’acquisition semble pouvoir être divisée par 10 par rapport à un protocole standard de
scanner de la cheville, sans que cela n’altère la qualité de l’analyse du mouvement. Ainsi,
même si une acquisition dynamique nécessite l’acquisition de 10 volumes, la dose globale
d’un scanner dynamique de la cheville pourrait être équivalente à la dose d’un scanner
standard de la cheville. Malheureusement, tandis qu’un premier test s’est révélé très concluant,
de nouveaux tests réalisés chez quatre autres cadavres n’ont pas permis d’obtenir les mêmes
résultats. Cela pourrait être lié à la qualité de l’os des cadavres. En effet, la trame osseuse des
quatre autres cadavres était très déminéralisée ce qui pourrait être à l’origine d’un défaut de
fonctionnement du logiciel de post-traitement (4D Ortho, Toshiba Medical System, Otawara,
Japon) utilisé pour quantifier les mouvements de la cheville. De nouveaux tests sur patients
seraient intéressants afin de poursuivre l’optimisation des scanners dynamiques de la cheville.

185
186
CONCLUSION GENERALE ET PERSPECTIVES

187
188
 1- Conclusion générale :

Dans cette thèse, nous avons étudié plusieurs facteurs techniques et comportementaux qui
permettent d’optimiser et de réduire la dose d’irradiation au scanner.

Nous avons montré que la sensibilisation des médecins prescripteurs de scanner est encore
à poursuivre compte tenu d’un défaut de connaissance de ces praticiens concernant les
niveaux de dose délivrée et les risques potentiels de cancer radio-induit lié aux faibles doses
de rayons X qui en découlent. Nous avons aussi montré que des mesures simples comme la
limitation du nombre de phases d’acquisition ou encore la réduction de la couverture
d’acquisition du scanner peuvent permettre chacune de réduire d’environ 20 % la dose
délivrée au patient dans certaines situations spécifiques. Souvent mise au second plan derrière
les innovations technologiques récentes, la mise en œuvre des facteurs comportementaux dans
une démarche d’optimisation et de réduction de la dose reste toutefois fondamentale.

Parallèlement, les évolutions technologiques récentes ont permis une réduction importante
de la dose d’irradiation au scanner. En premier lieu, l’implantation des reconstructions
itératives a permis de diviser par deux la dose d’irradiation au scanner par rapport aux
reconstructions standard en rétroprojection filtrée, à qualité d’image équivalente et sans
altération de la résolution spatiale des images. Nous avons aussi montré que pour des scanners
avec une faible longueur d’acquisition, l’utilisation du mode d’acquisition volumique à partir
d’un scanner à large système de détection permet de réduire la dose par rapport à l’acquisition
hélicoïdale classique. Enfin, l’utilisation de la modulation automatique du milliampérage
permet aussi d’optimiser la dose en adaptant le milliampérage au morphotype des patients et
en permettant d’avoir un bruit constant de l’image d’un patient à l’autre. Toutefois, dans notre
étude portant sur un protocole de scanner basse dose réalisé dans le cadre d’une suspicion de
colique néphrétique, nous avons montré que même si le bruit de l’image est constant d’un
patient à l’autre, la qualité d’image subjective et la performance diagnostique peuvent varier
en fonction du morphotype des patients. Ainsi, dans une démarche d’optimisation et de
réduction de la dose, la performance diagnostique est un critère d’évaluation plus pertinent
que la qualité d’image.

Enfin, nous avons proposé différentes manières de réduire et d’optimiser les doses en
pratique clinique courante. L’optimisation conjointe des facteurs techniques et
comportementaux permet la réalisation de scanner basse dose pour l’exploration des coliques

189
néphrétiques. Avec ce type de protocole, la dose peut être réduite de l’ordre de 70 à 90 %, par
rapport à une acquisition abdominopelvienne standard, sans altération de la performance
diagnostique. De même, la maitrise des doses par l’utilisation du mode d’acquisition
intermittent, par la limitation de la couverture d’acquisition et du nombre de phases
d’acquisition et par l’implantation des reconstructions itératives a permis l’utilisation en
pratique clinique courante des protocoles de scanner dynamique 4D des articulations ou
encore de scanner de perfusion tumorale.

2- Perspectives :

Durant la dernière décennie, les niveaux de dose d’irradiation au scanner ont été
largement diminués. Par exemple, pour un scanner abdominopelvien, grâce principalement à
l’implantation des reconstructions itératives mais aussi à l’amélioration des détecteurs et à
l’utilisation de la modulation automatique du milliampérage et de la collimation active, la
dose délivrée moyenne est passée de 10 mSv à environ 2,8 mSv [55], soit une réduction de
72 % de la dose, à qualité d’image équivalente.

Cette dynamique de réduction des doses délivrées au scanner ne semble pas s’arrêter. De
nouvelles générations de reconstruction itérative sont encore en cours de développement. Les
nouveaux algorithmes itératifs récemment commercialisés proposent d’ores-et-déjà des
niveaux de réduction de la dose de l’ordre de 70 à 80 % contre 30 à 50 % pour les algorithmes
de la première génération [65-66].

De nouvelles innovations technologiques sont aussi en cours de développement. Par

exemple, L’apparition récente des détecteurs compteurs de photons ou « Photon Counting
Detector » pourrait offrir de nouvelles opportunités de réduction de la dose d’irradiation au
scanner. Le principe de ce type de détecteur est d’utiliser un semi-conducteur ultra-rapide afin
de pouvoir compter directement chaque interaction avec un photon. Le remplacement des
détecteurs actuels à base de scintillateurs à cristaux par ces nouveaux détecteurs pourrait offrir
de nouveaux avantages comme la réduction de la sensibilité au bruit électronique,
l’augmentation du RCB, une amélioration de la résolution spatiale, une meilleure
décomposition des matériaux et enfin une réduction de la dose d’irradiation. [67]. Cette
nouvelle technologie est encore en cours d’évaluation mais des premiers tests sur des patients
ont montrés des résultats prometteurs [68].

190
 Un changement complet de technologie pourrait aussi intervenir et modifier totalement les
niveaux de dose d’irradiation délivrée en scanographie. Des chercheurs proposent d’introduire
une nouvelle technologie basée sur la détection ultrasonique de l’absorption des rayons X [69].
En effet, lors de l’interaction d’un photon X avec la matière, une onde ultrasonore est émise.
La détection de cette onde permet donc de savoir qu’il y a eu absorption d’un photon X. Des
résultats expérimentaux montrent que, si un tel système arrive à être développé pour une
application humaine, il serait possible d’avoir une vitesse d’acquisition des images cent fois
plus rapide tout en ayant une dose délivrée de rayons X cent fois plus faible [69]. Bien que
cette technologie ne soit encore qu’au stade de la recherche fondamentale, elle ouvre la porte
à de nombreuses perspectives pour les années à venir pour la réduction de la dose
d’irradiation au scanner.

Il serait aussi intéressant d’approfondir les recherches concernant l’impact des faibles
doses de rayons X sur le risque de cancer radio-induit. En effet, avec des doses de plus en plus
faibles, il pourrait s’avérer que leur impact devienne insignifiant voire nul, ce qui pourrait
alors changer totalement la manière d’utiliser le scanner en pratique clinique courante. De
même, de nos jours, les niveaux des doses délivrées au scanner s’approchent de plus en plus
des doses liées à la réalisation d’un bilan radiographique standard. C’est par exemple le cas
pour le scanner thoracique où des protocoles de scanner très basse dose ont montré une
performance diagnostique supérieure à la radiographie thoracique avec des doses d’irradiation
très proches [70]. Grâce à la réduction importante des doses délivrées en scanographie, les
indications de scanner pourraient alors être étendues à de nombreuses pathologies pour
lesquelles le bilan radiographique standard est de nos jours réalisé en première intention. Au
final, la réduction de plus en plus importante des doses délivrées au scanner permet de
supprimer son inconvénient majeur et offre au scanner de nouvelles et nombreuses
opportunités de développement futur.

191
192
REFERENCES BIBLIOGRAPHIQUES

1- Brenner DJ, Hall EJ. Computed tomography: an increasing source of radiation

exposure. N Engl J Med 2007; 357(22): 2277-84.

2- Etard C, Sinno-Tellier S, Aubert B. Exposition de la population française aux

rayonnements ionisants liée aux actes de diagnostic médical en 2007. Rapport
InVS/IRSN. Juin 2010.

3- Exposition de la population française aux rayonnements ionisants liée aux actes de

diagnostic médical en 2012. Rapport PRP-HOM N°2014-6.

4- Board of Radiation Effects Research Division on Earth and Life Sciences National
Research Council of the National Academies. Health risks from exposure to low levels
of ionizing radiation: BEIR VII Phase2. Washington, DC: National Academies Press;
2006.

5- Brenner DJ, Doll R, Goodhead DT, Hall EJ, Land CE, Little JB, et al. Cancer risks
attributable to low doses of ionizing radiation: assessing what we really know. Proc
Natl Acad Sci USA 2003; 100(24): 13761-6.

6- Little MP, Wakeford R, Tawn EJ, Bouffler SD, Berrington de Gonzalez A. Risks
associated with low doses and low dose rates of ionizing radiation: why linearity may
be (almost) the best we can do. Radiology 2009; 251(1): 6-12.

7- Smith-Bindman R, Lipson J, Marcus R, Kim KP, Mahseh M, Gould R, et al.

Radiation dose associated with common computed tomography examinations and the
associated lifetime attributable risk of cancer. Arch Intern Med 2009; 169(22): 2078-
86.

8- Tubiana M, Aurengo A, Averbeck A. La relation dose-effet et l’estimation des effets

cancérigènes des faibles doses de rayonnements ionisants. Paris : Académie Nationale
de Médecine, Institut de France – Académie des Sciences; 2005.

9- Tubiana M, Feinendegen LE, Yang C, Kaminski JM. The linear no-threshold

relationship is inconsistent with radiation biologic and experimental data. Radiology
2009; 251(1): 13-22.

193
10- Doss M. Linear no-threshold model vs radiation hormesis. Dose Response 2013; 11:
480-97.

11- Directive 97/43/Euratom du 30 juin 1997 relative à la protection sanitaire des

personnes contre les dangers des rayonnements ionisants lors d’expositions à des fins
médicales.

12- Directive 2013/59/Euratom du 5 décembre 2013 fixant les normes de base relatives à
la protection sanitaire contre les dangers résultant de l’exposition aux rayonnements
ionisants.

13- Wallace AB, Goergen SK, Schick D, Soblusky T, Jolley D. Multidetector CT dose:
clinical practice improvement strategies from a successful optimization program. J Am
Coll Radiol 2010; 7(8): 614-24.

14- Tamm EP, Rong XJ, Cody DD, Ernst RD, Fitzgerald NE, Kundra V. Quality
initiatives: CT radiation dose reduction: How to implement change without sacrificing
diagnostic quality. Radiographics 2011; 31(8): 1823-32.

15- Amis ES Jr. CT radiation dose: trending in the right direction. Radiology 2011;
261(1): 5-8.

16- Brenner DJ, Shuryak I, Einstein AJ. Impact of reduced patient life expectancy on
potential cancer risks from radiologic imaging. Radiology 2011; 261(1): 193-8.

17- Oikarinen H, Meriläinen S, Pääkkö E, Karttunen A, Nieminen MT, Tervonen O.

Unjustified CT examinations in young patients. Eur Radiol 2009; 19(5): 1161-5.

18- Chuong AM, Corno L, Beaussier H, Boulay-Coletta I, Millet I, Hodel J, et al.

Assessment of bowel wall enhancement for the diagnosis of intestinal ischemia in
patients with small bowel obstruction. Radiology 2016 Feb 11: 151029 [Epub ahead of
print].

19- Kurabayashi M, Okishige K, Ueshima D, Yoshimura K, Shimura T, Suzuki H, et al.

Diagnostic utility of unenhanced computed tomography for acute aortic syndrome.
Circ J 2014; 78(8): 1928-34.

194
20- Esposito AA, Zilocchi M, Fasani P, Giannitto C, Maccagnoni S, Maniglio M, et al.
The value of precontrast thoraco-abdominopelvic CT in polytrauma patients. Eur J
Radiol 2015; 84(6): 1212-8.

21- Badawy MK, Galea M, Mong KS, U P. Computed tomography overexposure as a

consequence of extended scan length. J Med Imaging Radiat Oncol 2015; 59(5): 586-9.

22- Michalakis N, Keyzer C, De Maertelaer V, Tack D, Genevois PA. Reduced z-axis

coverage in multidetector-row CT pulmonary angiography decreases radiation dose
and diagnostic accuracy of alternative diseases. Br J Radiol 2014; 87(1033): 20130546.

23- Atalay MK, Walle NL, Grand DJ, Mayo-Smith WW, Cronan JJ, Egglin TK. Scan
length optimization for pulmonary embolism at CT angiography: analysis based on the
three-dimensional spatial distribution of 370 emboli in 100 patients. Clin Radiol 2011;
66(5): 405-11.

24- Kalra MK, Toth TL. Patient centering in MDCT: dose effects. In: Tack D, Genevois
PA (eds) Radiation dose from adult and pediatric multidetector computed tomography.
Springer, Berlin, 2007, pp 129-132.

25- Li J, Udayasankar UK, Toth TL, Seamans J, Small WC, Kalra MK. Automatic patient
centering for MDCT: effect on radiation dose. Am J Roentgenol 2007; 188(2): 547-52.

26- Mahesh M. Scan parameters and image quality in MDCT. In: Mahesh M (eds) MDCT
physics: The basics—Technology, image quality and radiation dose. Lippincott
Williams & Wilkins, Philadelphia, 2009, pp 47-78.

27- Ciarmatori A, Nocetti L, Mistretta G, Zambelli G, Costi T. Reducing absorbed dose to

eye lenses in head CT examinations: the effect of bismuth shielding. Australas Phys
Eng Sci Med 2016 Apr 20. [Epub ahead of print].

28- van der Molen AJ, Geleijns J. Overranging in multisection CT: quantification and
relative contribution to dose- comparison of four 16-section CT scanners. Radiology
2007; 242(1): 208-16.

195
29- Christner JA, Zavaletta VA, Eusemann CD, Walz-Flannigan AI, McCollough CH.
Dose reduction in helical CT: dynamically adjustable z-axis X-ray beam collimation.
Am J Roentgenol 2010; 194(1): W49-55.

30- Boedeker KL, McNitt-Gray MF. Application of the noise power spectrum in modern
diagnostic MDCT: part II. Noise power spectra and signal to noise. Phys Med Biol
2007; 52(14): 4047-61.

31- Kalra MK. Automatic exposure control in multidetector-row CT. In: Tack D,
Genevois PA, Kalra M. (eds). Radiation dose from multidetector CT. Springer-Verlag
Berlin Heidelberg, 2012, p369-387.

32- Singh S, Kalra MK, Thrall JH, Mahesh M. Automatic exposure control in CT:
applications and limitations. J Am Coll Radiol 2011; 8(6): 446-9.

33- Kalender WA, Deak P, Kellermeier M, van Straten M, Vollmar SV. Application and
patient size-dependent optimization of x-ray spectra for CT. Med Phys 2009; 36(3):
993-1007.

34- Nagel HD. CT parameters that influence the radiation dose. In: Tack D, Genevois PA
(eds) Radiation dose from adult and pediatric multidetector computed tomography.
Springer, Berlin, 2007, pp 51-79.

35- Kalra MK, Maher MM, Toth TL, Hamberg LM, Blake MA, Shepard JA, et al.
Strategies for CT radiation dose optimization. Radiology 2004; 230(3): 619-28.

36- Stradiotti P, Curti A, Castellazzi G, Zerbi A. Metal-related artifacts in instrumented

spine. Techniques for reducing artifacts in CT and MRI: state of the art. Eur Spine J
2009; 18 Suppl 1: 102-8.

37- von Falck C, Galanski M, Shin HO. Sliding-thin-slab averaging for improved
depiction of low-contrast lesions with radiation dose savings at thin-section CT.
Radiographics 2010; 30(2): 317-26.

38- McNitt-Gray MF. AAPM/RSNA physics tutorial for residents: topics in CT. Radiation
dose in CT. Radiographics 2002; 22(6): 1541-53.

196
39- Walter F, Ludig T, Iochum S, Blum A. Multi-dectector CT in musculo-skeletal
disorders. JBR-BTR 2003; 86(1): 6-11.

40- Blum A. Scanner volumique multicoupe: principes, applications et perspectives. JBR-

BTR 2002; 85(2): 82-99.

41- Xu J, Mahesh M, Tsui BM. Is iterative reconstruction ready for MDCT? J Am Coll
Radiol 2009; 6(4): 274-6.

42- Thibault JB, Sauer KD, Bouman CA, Hsieh J. A three-dimensional statistical approach
to improved image quality for multislice helical CT. Med Phys 2007; 34(11): 4526-44.

43- Hara AK, Paden RG, Silva AC, Kujak JL, Lawder HJ, Pavlicek W. Iterative
reconstruction technique for reducing body radiation dose at CT: feasibility study. Am
J Roentgenol 2009; 193(3): 764-71.

44- Krille L, Hammer GP, Merzenich H, Zeeb H. Systematic review on physician’s

knowledge about radiation doses and radiation risks of computed tomography. Eur J
Radiol 2010; 76(1): 36-41.

45- Brown N, Jones L. Knowledge of medical imaging radiation dose and risk among
doctors. J Med Imaging Radiat Oncol 2013; 57(1): 8-14.

46- Corwin MT, Bekele W, Lamba R. Bony landmarks on computed tomography localizer
radiographs to prescribe a reduced scan range in patients undergoing multidetector
computed tomography for suspected urolithiasis. J Comput Assist Tomogr 2014;
38(3): 404-7.

47- Corwin MT, Chang M, Fananapazir G, Seibert A, Lamba R. Accuracy and radiation
dose reduction a limited abdominopelvic CT in the diagnosis of acute appendicitis.
Abdom Imaging 2015; 40(5): 1177-82.

48- Kroft LJ, Roelofs JJ, Geleijns J. Scan time and patient dose for thoracic imaging in
neonates and small children using axial volumetric 320-detector row CT compared to
helical 64-, 32-, and 16- detector row CT acquisitions. Pediatr Radiol 2010; 40(3):
294-300.

197
49- Zacharias C, Alessio AM, Otto RK, Iyer RS, Philips GS, Swanson JO, et al. Pediatric
CT: strategies to lower radiation dose. Am J Roentgenol 2013; 200(5); 950-6.

50- Semelka RC, Armao DM, Elias J, Huda W. Imaging strategies to reduce the risk of
radiation in CT studies, including selective substitution with MRI. J Magn Reson
Imaging 2007; 25(5): 900-9.

51- McCollough CH, Primak AN, Braun N, Kofler J, Yu L, Christner J. Strategies for
reducing radiation dose in CT. Radiol Clin North Am 2009; 47(1): 27-40.

52- Lee TY, Chhem RK. Impact of new technologies on dose reduction in CT. Eur J
Radiol 2010; 76(1): 28-35.

53- Singh S, Kalra MK, Thrall JH, Mahesh M. CT radiation dose reduction by modifying
primary factors. J Am Coll Radiol 2011; 8(5): 369-72.

54- Dougeni E, Faulkner K, Panayiotakis G. A review of patient dose and optimisation

methods in adult and paediatric CT scanning. Eur J Radiol 2012; 81(4): e665-83.

55- McCollough CH, Chen GH, Kalender W, Leng S, Samei E, Taguchi LK, et al.
Achieving routine submillisievert CT scanning: report from the summit on
management of radiation dose in CT. Radiology 2012; 264(2): 567-80.

56- Coursey CA, Casalino DD, Remer EM, Arellano RS, Bishoff JT, Dighe M, et al. ACR
Appropriateness Criteria® acute onset flank pain: suspicion of stone disease.
Ultrasound Q 2012; 28(3): 227-33.

57- Kambadakone AR, Eisner BH, Catalano OA, Sahani DV. New and evolving concepts
in the imaging and management of urolithiasis: urologists’ perspective. Radiographics
2010; 30(3): 603-23.

58- Broder J, Bowen J, Lohr J, Babcock A, Yoon J. Cumulative CT exposures in

emergency department patients evaluated for suspected renal colic. J Emerg Med
2007; 33(2): 161-8.

59- Katz SI, Saluja S, Brink JA, Forman HP. Radiation dose associated with unenhanced
CT for suspected renal colic: impact of repetitive studies. Am J Roentgenol 2006;
186(4): 1120-4.

198
60- Niemann T, Kollmann T, Bongartz G. Diagnostic performance of low-dose CT for
detection of urolithiasis: a meta-analysis. Am J Roentgenol 2008; 191(2): 396-401.

61- Sung MK, Singh S, Kalra MK. Current status of low dose multi-detector CT in the
urinary tract. World J Radiol 2011; 3(11): 256-65.

62- Glazer DI, Maturen KE, Cohan RH, Davenport MS, Ellis JH, Knoepp US, et al.
Assessment of 1 mSv urinary tract stone CT with model-based iterative
reconstruction. Am J Roentgenol 2014; 203(6): 1230-5.

63- McLaughlin PD, Murphy KP, Hayes SA, Carey K, Sammon J, Crush L, et al. Non-
contrast CT at comparable dose to an abdominal radiograph in patients with acute
renal colic; impact of iterative reconstruction on image quality and diagnostic
performance. Insights Imaging 2014; 5(2): 217-30.

64- Biswas D, Bible JE, Bohan M, Simpson AK, Whang PG, Grauer JN. Radiation
exposure from musculoskeletal computerized tomographic scans. J Bone Joint Surg
Am 2009; 91(8): 1882-9.

65- Dodge CT, Tamm EP, Cody DD, Liu X, Jensen CT, Wei W, et al. Performance
evaluation of iterative reconstruction algorithms for achieving CT radiation dose
reduction - a phantom study. J Appl Clin Med Phys 2016; 17(2): 5709.

66- Lubner MG, Pickhardt PJ, Tang J, Chen GH. Reduced image noise at low-dose
multidetector CT of the abdomen with prior image constrained compressed during
sensing algorithm. Radiology 2011; 260(1): 248-56.

67- Taguchi K, Iwanczyk JS. Vision 20/20: Single photon counting x-ray detectors in
medical imaging. Med Phys 2013; 40(10): 100901.

68- Pourmorteza A, Symons R, Sandfort V, Mallek M, Fuld MK, Henderson G, et al.

Abdominal imaging with contrast-enhanced photon-counting CT: first human
experience. Radiology 2016; 279(1):239-45.

69- Xiang L, Han B, Carpenter C, Pratx G, Kuang Y, Xing L. X-ray acoustic computed
tomography with pulsed x-ray beam from a medical linear accelerator. Med Phys
2013; 40(1): 010701.

199
70- McLaughlin PD, Ouellette HA, Louis LJ, Mallinson PI, O-Connell T, Mayo JR, et al.
The emergence of ultra-low-dose computed tomography and the impending
obsolescence of the plain radiograph? Can Assoc Radiol J 2013; 64(4):314-8.

200
 RESUME

Depuis son introduction dans les années 1970, le scanner est devenu une technique d’imagerie
médicale incontournable grâce à son excellente performance pour le diagnostic de
nombreuses pathologies. Toutefois, le scanner est un examen d’imagerie irradiant. Compte-
tenu des risques potentiels de cancer radio-induit liés aux faibles doses de rayons X, la
réduction de la dose d’irradiation au scanner est primordiale. Dans ce travail, nous avons
étudié plusieurs facteurs techniques et comportementaux qui permettent d’optimiser et de
réduire la dose d’irradiation au scanner, tout en préservant une excellente performance
diagnostique. Du côté des facteurs comportementaux, la sensibilisation des équipes médicales
et paramédicales est fondamentale dans une démarche d’optimisation de la dose d’irradiation
au scanner. De même, la limitation du nombre de phases d’acquisition et la réduction de la
couverture d’acquisition sont deux manières simples pour réduire les doses délivrées. Du côté
des facteurs techniques, nous avons montrés que l’utilisation des reconstructions itératives,
par rapport aux reconstructions standards en rétroprojection filtrée, permet de réduire de
moitié la dose d’irradiation des scanners, à qualité d’image équivalente. L’acquisition en
mode volumique pour les scanners avec une faible couverture d’acquisition et l’utilisation de
la modulation automatique du milliampérage permettent aussi de réduire et d’optimiser les
doses. Enfin, nous nous sommes intéressés à l’optimisation de protocoles de scanner en
pratique clinique courante en se focalisant sur les scanners réalisés pour la recherche d’une
colique néphrétique et pour les scanners en imagerie ostéo-articulaire. Dans ce dernier
domaine, nous avons aussi proposé des protocoles de scanner pour des applications cliniques
avancées comme le scanner dynamique des articulations ou le scanner de perfusion tumorale.

Mots-clés : Dose d’irradiation ; Optimisation ; Rayons X ; Réduction ; Scanner.

ABSTRACT
Since its introduction in the 1970s, computed tomography (CT) has become the technique of
reference in medical imaging for many diseases due to its high diagnostic performance. Its
main limitation is the radiation dose delivered to the patient. Considering the potential risks of
radiation-induced cancer caused even with low dose exposure, dose reduction in CT is
essential. In this work, we studied several technical and behavioral factors that allow for CT
radiation dose reduction and optimization, without modifying the diagnostic performance.
Among the behavioral factors studied, education and awareness of radiologists and radiology
technicians are important elements for CT radiation dose reduction. Limiting CT scan
coverage and the number of acquisition phases is also a straightforward and effective way to
reduce dose exposure. Regarding technical factors, we have shown that iterative
reconstruction algorithms can reduce in half the radiation dose in comparison with standard
filtered back projection, while maintaining equivalent image quality. The use of wide volume
mode for acquisitions with a short coverage and the use of the automatic tube current
modulation can also be used to reduce and optimize CT radiation dose. Finally, we provide
guidelines to optimize CT radiation dose in some clinical settings such as renal colic and
musculoskeletal CT. We also propose practical guidelines for advanced clinical applications
of joint dynamic CT and perfusion CT in musculoskeletal disease.

Keywords: Computed Tomography; Optimization; Radiation dose; Reduction; X-Ray.