Computed and Digital Radiography

Digital Imaging Initial Processing

PREPROCESSING
• Refers to the initial steps being applied to the raw imaging
data, which is beyond the control of the radiologic
technologist
• Mainly depends on the readout characteristics of the
detector being used in the CR/DR system

STEPS IN IMAGE ANALYSIS

• Detect collimation mask
• Detect direct exposure
• Extract anatomy regions
• Calculate key image descriptors

IMAGE ANALYSIS
• EXPOSURE FIELD RECOGNITION
o Processes only the optimal density exposure range
from the wide range of exposure received by the
detector
o Includes the determination of the collimation edges
DETECTOR SCANNING CHARACTERISTICS
and eliminating the scatter density values outside the
• POINT SCAN DETECTOR SYSTEM
collimated area
o Laser-based CR
o Simplest digital scanning technique
o Uses a single-pixel cell detector to scan an image
sequentially throughout a series of X and Y
coordinates
o More prone to registration errors
• LINEAR SCAN DETECTOR SYSTEM
o Newer CR and some charged-couple devices
o Higher image acquisition rate than point scan system
o Linear scanning along a single axis to produce an
image
• 2D FIXED ARRAY SYSTEM
o CCD, indirect and direct DR • SIGNAL SAMPLING
o Most sophisticated digital scanning technique o The Nyquist theorem (Nyquist-Shannon theorem,
o Elimination of movement of the image sensor sampling theorem) states that when sampling a signal
o Despite having the fastest acquisition rate and high (converted from analog image to a digital image), the
degree of accuracy between pixels, the use of such sampling frequency must be greater than twice the
produces lower signal to noise ratio (SNR) than the frequency of the input signal so that the
one-dimensional fixed array system reconstruction of the original image will be as close to
the original signal as possible
o If too few pixels are sampled, the result will be lack of
resolution or commonly known as aliasing
✓ Aliasing – results in appearance of more effect
(odd stripes and patterns)

• SEGMENTATION
o After digitization, these signals must be divided into
their respective constituent regions or objects within
the potential digital image
o Involves the partition of image pixels into groups that
strongly correlate with the object in the image
o Segmentation algorithms are based on one of two
basic properties of intensity values
✓ SIMILARITY – partitioning of an image into regions
that are similar according to a set of predefined
criteria
 Computed and Digital Radiography
✓ DISCONTINUITY – partitioning of an image into
regions based on abrupt changes in intensity
values

• LOOKUP TABLE
o Provide a method of processing digital images in order
• HISTOGRAM FORMATION to change the brightness and contrast displayed using
o Histogram – the graphical representation of all the computer algorithms
digitized intensities recorded on the detector, with the o Because digital IRs have a linear exposure response
x-axis representing the amount of exposure read, and and a very large dynamic range, raw data images
y-axis representing the number of pixels for each exhibit low contrast and must be altered to improve
exposure visibility of anatomic structures
o Lower energy (low kVp) = wider histogram o Look up tables are also sometime used to reverse or
o High energy (high kVp) = narrower histogram invert image grayscale
o Left side of the graph = lower signal values (less
exposure, black in color)
o Right side = higher signal values (more exposure, white
in color)

• HISTOGRAM ANALYSIS
o An image processing technique commonly used to
identify the edges of the image and assess the raw
data prior to image display. In this method, the Photo A visually compares pixel values of the original
computer first creates a histogram of the image image with a processed image. If the image is not altered, the
o Computer analyzes the histogram using processing graph would be a straight line. If the original image is altered,
algorithms and compares it with a pre-established the original pixel values would be different in the processed
histogram specific to the anatomic part being imaged image and the graph would no longer be a straight line but
o The stored histogram models have value of interest might resemble a characteristic curve for radiographic film
(VOI) which determine the range pf the histogram data (Photo B)
set that should be included in the displayed image

Lookup table altering the pixel values of a low-contrast

image to display an image with higher contrast.

• AUTOMATIC RESCALING
o It is employed to maintain consistent image
brightness despite overexposure or underexposure
✓ Dose Creep – refers to the potential gradually
increase patient exposure over time

The original low-contrast chest image is altered to display

a higher-contrast chest image. The graph shows a change in the
pixel values from the original image.
 Computed and Digital Radiography
• SPATIAL FREQUENCY FILTERING
o EDGE ENHANCEMENT
✓ High pass filtering
✓ Occurs when fewer pixels around are involved in
the averaging
✓ The smaller the neighborhood, the greater the
enhancement is
✓ Noted increase with both the contrast and edge
enhancement for the reason that the frequencies
of areas of interest are amplified while
suppressing the other frequencies (masking).
However, masking can result to loss of small
details
✓ Useful for enhancement of large structures such
as soft tissues but it could be noisy
✓ Over-processing of the image lead to appearance
of halo effect

o SMOOTHING
✓ Low pass filtering
✓ Involves averaging each pixel’s frequency with
surrounding pixel to remove high-frequency
noise
✓ Useful for viewing small details such as fine bone
tissue

o SIGNAL EQUALIZATION
✓ Automatically accommodate changes in latitude
✓ Compresses the image signal dynamic range such
that all information within that region of interest
can be rendered with optimal contrast