A PRECISE METHOD FOR MEASURING
LUMBAR SPONDYLOLISTHESIS
Project Work
Phase-II
Final Review
ME CSE 1
OBJECTIVES:
Lumbar spondylolisthesis is one of the
most common spinal diseases. It is caused
by the anterior shift of a lumbar vertebrae
relative to its subjacent vertebrae. The
manual measurement of the relative shift
between neighboring lumbar vertebrae ,
which is time consuming and
irreproducible. Thus, an automatic
algorithm to measure lumbar vertebrae
shift becomes desirable for
spondylolisthesis diagnosis and staging .
BLOCK DIAGRAM
EXISTING SYSTEM:
• Automatic localization and
identification of vertebrae in
arbitrary field-of-view CT scans. No
assumptions are made about which
section of the spine is visible or to
which extent.
DRAWBACKS OF EXISTING SYSTEM:
Due to the small size of the vertebrae,
slight errors of detection may lead to
significant measurement errors, hence,
wrong disease stages.
• Automatic localize and label each lumbar
vertebrae is required to provide the
semantic meaning of the measurement. It
is difficult since different lumbar
vertebrae have high similarity of both
shape and image appearance.
PROPOSED SYSTEM:
• A learning based spine labeling method
that integrates both the image
appearance and spine geometric
information is used for detection of each
lumbar vertebrae.
• A hierarchical method which uses both
the population information from atlases
and domain-specific information in the
target image is proposed for most anterior
and posterior points positioning
PROPOSED ALGORITHM:
• CARP(Critical Anatomy Region
Propagation)
New framework for automatic
spondylolisthesis measurement
PROPOSED SYSTEM ADVANTAGES:
• This method is efficient, as it takes
only several seconds in average to
estimate the spondylolisthesis grade
of each lumbar vertebrae pair for
one CT spine image.
APPLICATION:
• Spinal diseases applications
MODULE:
• AUTOMATIC SPINE LABELING WITH HIERARCHICAL
LEARNING
• COURSE LEVEL REPRESENTATION
• FINE LEVEL REPRESENTATION
Phase-II Implementation
Future research includes to train the vertebrae
detectors using random forests to improve the
robustness of detection instead of AdaBoost
Classification
Random Forest
Random forests or random decision forests are an
ensemble learning method for classification,
regression and other tasks, that operate by
constructing a multitude of decision trees at training
time and outputting the class that is the mode of the
classes (classification) or mean prediction
(regression) of the individual trees.
Input Image
Automatic Spine Labeling with Hierarchical
Learning
Spine Labeling
Course Level Analysis
CARP –Critical Anatomy Region
Propagation
Output
Fine Level Analysis
Disease Detection Input Image
Automatic Spine Labeling with Hierarchical
Learning
Spine Labeling
Course Level Analysis
Course Level Analysis
CARP-Critical Anatomy Region Propagation
Output
Fine Level Analysis
Hardware Requirements:
• Processor : core I3 Processor
• Hard Disk : 500GB
• RAM : 4GB (minimum)
• Keyboard : 110 keys enhanced
Software Requirements:
MATLAB 8.6 2015b
FUTURE ENHANCEMENT:
• Future research includes to train
the vertebrae detectors using deep
learning or random forests to
improve the robustness of the
detection.