UNIT 3: Digital Imaging Pre-Processing Flashcards
Pre-processing
performed by the computer on the raw image data to correct for flaws in the
image acquisition process (acquisition processing); largely automatic
Post-processing
all manipulation and adjustments made in order to refine the image
Pre-processing vs. Post-processing
-Pre-processing: image corrections
-Post-processing: image refinement
Digital Imaging Preprocessing Problems and Solutions
-Defective Pixel: Interpolate adjacent pixel signal
-Image Lag: Offset voltage correction
-Line Noise: Correct from dark reference zone
How are post-processing functions and operator adjustments related?
-Post-processing: all manipulation and adjustments made in order to refine the image (includes default and operator adjustments)
-Operator Adjustments: Post-processing functions carried out by the user after initial display
Post-Processing examples
(also image zoom)
5 steps in pre-processing
- Field uniformity
- Noise reduction for DEL dropout
- Exposure field recognition
- Histogram formation and Analysis
- Histogram rescaling
Post-Processing steps
- Gradation Processing (LUT)
- Detail Processing
- Formatting for Display
- Image Displayed
- Operator Adjustments
Field Uniformity potential sources of noise in image acquisition
-Faulty DELs/ DEL drop out
-Dark noise/Ambient Nose in CR (Background radiation)
-Dark current in DR and CR
-Phosphor layer thickness variations for indirect capture DR and CR systems
-Light guide variations in CR
-Anode heel effect especially at short SID’s
What does Field uniformity do?
correct for flaws in image acquisition
Field uniformity: Faulty DEL’s
Field uniformity: Dark Noise
-In CR
-Accumulation of background radiation on the CR plate prior to/or after the image is acquired
-Ensure the IR is erased prior to use and processed immediately after exposure
Field uniformity: Dark Current
-In DR and CR
-(noise) a small amount of current remaining in the electronics in the absence of x-ray exposure
Dark current in DR vs CR
-DR: caused from current moving in the electronics of the detector in the absence of x-ray exposure
-CR: typically is attributed to the CCD but may be caused by light guide variations in the reader as well.
(These elements create noise in the system that can degrade image quality and efficiency of the system)
Does DQE increase or decrease as Dark Current increases?
Decrease
What is a common QC test performed by radiographers on equipment to test for field uniformity issues?
-Flat fielding AKA Gain Calibration or Calibration tests
-A QC test performed on a digital detector by producing an image on the receptor without any attenuating material in the field of view
Flat field uniformity tests
-In image A, a flat field uniformity test demonstrates visual artifacts from the Bucky tray mechanism on a digital mammography machine.
-In image B, a DR image picked up the ionization chambers inside the Bucky tray. These artifacts must be detected and corrected before clinical use of digital equipment
Step 2: Noise reduction for DEL dropout. What is DEL dropout?
a DEL may be dead or faulty causing noise and/or signal loss. This can be a single DEL or an entire row.
Step 2 Noise Reduction for DEL dropout. What is a kernel? What is Interpolation?
-Kernel: (A matrix within a matrix) A computer processing code that is applied to an entire data set
-Interpolation: average of surrounding cells to fill in the gap
Dead DELS or Pixels: Kernels and Interpolation (malfunctioning pixels/detector elements)
To correct for a non-uniform field in which a small area of the detector is not picking up a signal (AKA dead DEL or pixel), a kernel is configured to average the signals surrounding the malfunctioning DEL and fill in the dead space with interpolation
Step 3: Exposure Field Recognition. What is another term for this?
Edge Detection Recognition
Step 3: Exposure Field Recognition
• System excludes areas outside the collimation from processing
-Requires parallel and symmetric collimation
Step 3: Exposure Field Recognition. What happens when it fails?
-This may lead to histogram errors resulting in issues with poor resolution, overly light or dark images and/or EI numbers that do not accurately represent the exposure to the IR
-Reasons for EFR/EDR failure:
• High exposure to the detector
• Extreme off centering of part on IR
• Improper collimation
• Area of high attenuation within the image (metal objects, barium, lead shield)
Step 3: Exposure Field Recognition. What is segmentation software?
-Creating multiple exposures on 1 IR
-Used in CR ONLY to identify and count the number of views per image receptor
Step 4: Histogram Formation and Analysis. What is a histogram?
A graphic representation of the gray shades within an image
Parts of a histogram
A. Threshold
B. Smin
C. VOI
D. Smax
E. Tail
Histogram Analysis Type 1
-Expects a spike for Direct Exposure (High density objects such as bone, attenuate the beam more so than air. The area around a hand exposure doesn’t get attenuated and so there is direct exposure to the IR in those areas)
-Smax will not include the spike (tail)
-Ex: Extremity work with a collimated field
Histogram Analysis Type 2
-Assumes there will be no spike for Direct Exposure
-Ex: Abdomen work
-Includes everything from the threshold value to the darkest registered value in the VOI
Histogram Type 3
-Expects a spike for high density object (and sometimes a spike for Direct Exposure)
-Ex: Barium Studies
-Excludes BOTH spikes from the VOI
Grayscale Curves (Look Up Table): Long Scale Contrast vs Short Scale Contrast
How do technologists cause histogram analysis error?
• Selecting improper exam type
• Poor collimation
• Poor patient positioning or centering
• Objects within the exposure field
- Manual reprocessing of the histogram may be allowed, but image may need to be repeated
Step 5: Histogram Rescaling. What is rescaling?
The basic set of default algorithms applied to the raw data to make the image appear within human visual limits or “normal”
*corrects for over and under exposure(has to do with brightness)
Step 5: Histogram Rescaling. What is Normalization/Equalization?
-Initial processing algorithms applied to image data to modify image brightness
•Computer compares acquired image histogram with reference image histogram and adjusts for overall image brightness
What does rescaling correct the image for?
Under exposure and Over exposure (rescaling has to do with brightness)
Step 5: Histogram Rescaling. S and Q values
-Rescaling uses a Table to assign new output values to incoming pixel values
•S values = input values (new image)
•Q values= output values (reference histogram)
Rescaling vs Window Level
-Rescaling: pre-processing
-Window Level: post-processing
Define default processing
Operations that the computer does automatically without user input. These include both preprocessing and display or presentation processing.
Define presentation or display processing
The initial preparation or “shaping” of the digital data for viewing. These steps are performed to bring the data into human visual range so radiologists can interpret. These processes are part of the default processing operations performed by the computer.
Default Processing steps
- latent digital image acquired
- flat fielding corrections for field uniformity
- Electronic noise reduction
- signal drop out corrections
- exposure field recognition
- brightness processing (histogram)
- gradation processing (LUT)
- detail processing
- formatting for display
- image displayed
Define Interpolation
An operation that estimates an unknown value from 2 or more known values. In other words, the surrounding DELs are sampled and averaged and the result is inserted as the value for the dead DEL.
What is Image Lag and what causes it?
-Residual electronic signal left in the electronic channels of the TFT after exposure.
-This can be caused by:
• rapid exposures; i.e., not enough time b/t exposures for complete erasure to take place
• overexposure to the detector
• poor collimation: large areas of unattenuated beam on the detector
Image lag is more noticeable
when switching from high to low dose techniques
Best solution for Dark Noise is
to ensure the IP(Image Receptor) is erased just before using it to expose a patient.
Prior to any manipulations of the image data, such as calculating the exposure indicator or rescaling the image for brightness or contrast, the digital image characteristics and components must be counted and analyzed. How is this done?
The computer does this though processes called pattern separation or segmentation and exposure field recognition
Partitioned Pattern Recognition software (aka Segmentation)
looks for the abrupt changes in density between very dark exposure to very light.
Define Masking aka Shuttering or Cropping
-Post processing feature that can improve image visibility by removing the white area surrounding the image that may produce veiling glare
-The act of applying a black border to eliminate the white areas around a properly collimated image
Define Veiling glare
A phenomenon that occurs when too much ambient light scatters around an object making the image appear “washed out”
T or F: For ALARA and image quality, masking should never be used in place of good collimation
True
Masking is also called
Shuttering and Cropping
Does LUT have to do with brightness or contrast?
Contrast
Histogram rescaling changes
the image brightness to match the digital reference image brightness.
What is the Savg?
Affects exposure Indicator, average exposure to the image receptor
