Digital Image Pre-Processing Flashcards
What is pre-processing?
Corrections made to raw image data due to physical flaws in image acquisition
What is post-processing?
Image manipulation after initial corrections, based on personal preference
What is a common misconception and pre and post image processing?
People often think these happen sequentially, but they involve similar image correction actions
When does pre-processing occur?
Immediately after the raw image is captured and before the final product is displayed on screen
How does pre-processing work?
The system will scan the image for errors and correct them based on histograms
What are histograms?
Pre-programmed graphs that indicate all brightness levels of the pixels in an image that are used for pre-processing comparison
What will the system do if differences are noted between the Lookup-Table and exposure histogram?
It will attempt to reconcile the exposure histogram against the lookuptable and edit it to match
Why would a tech complete post-processing on an image that was already pre-processed?
- Personal preference
- Correct any system issues that might have happened incorrecly in pre-processing
What are two types of correction completed in pre-processing?
Flat field uniformity
Dead pixels
What is flat field uniformity?
A corrective feature designed to even out the brightness across an image
How is flat field uniformity tested for?
By examining the pixels across an exposure field in different corners/center of the picture to make sure they’re of equal brightness
If the brightness is not the same across the exposure field, what could that be an indication of?
Machine sensitivity is off
What are dead pixels?
Black pixels in an image that are called drop-outs
What causes dead pixels?
TFT failure in CR/DR systems
How can dead pixels be corrected?
Using a process called interpolation where an average of 8 pixels surrounding the dead pixel are captured and used to fill in the blank space with appropriate gray scale
What is the area surrounding the dead pixels called?
Kernal
What are the image analysis components completed in pre-processing?
- Segmentation
- Partioned pattern recognition (CR only)
- Exposure field recognition
- Construction of initial histogram
- Histogram analysis
What is segmentation?
The identification of areas of noise and associated densities outside the collimated field
What is partioned pattern recognition?
Identification of separate collimation fields only used in CR
What is a common error of PPR?
Rescaling error where the borders of collimation are not accurately recognized which affects the image brightness
What is scaling?
The process of creating the optimal contrast in an image and changing it if it doesn’t match
What is exposure field recognition?
Analysis of background densities out the anatomy by looking at anatomical landmarks called Volumes of Interest (VOI)
What are the types of landmark changes being looked at in exposure field recognition?
Changes between bone, soft tissue and metal
What is poor field recognition caused by?
A rescaling issue when the computer doesn’t understand the EFR even though techniques are set correctly
What are some instances when poor field recognition can occur?
Bilateral hands when both hands are imaged together or when lead squares are used for lateral spines
What does a histogram identify?
All densities (brightness levels) on the imaging plate in the form of a graph
What doe the x axis of a histogram represent?
The amount of exposures
What does the y axis of a histogram respresent?
The number of pixels for each expoosure
What is reflected on the left side of a histogram?
Bright areass (bone, contrast or metal)
What is reflected on the right side of a histogram?
Dark areas (air/gas)
How is a histogram usually assessed?
With a starting point (Smin) and a End point (Smax)
What do the spikes at the end of a histogram after the Smax reflect?
Raw background radiation outside the image
What are the 3 types of histograms?
Type 1 = Less than full IR
Type 2: Full IR
Type 3: Barium and shield
What are the properties of a Type 1 Histogram?
Histogram with one spike of radiation, due to the anatomy not covering the entire IR
What does the spike at the end of a Type 1 histogram represent?
Background radiation outside the anatomy of interest
When moving from left to right on a Type 1 hisotgram how does the density change?
Density gets less dense and darker as you move right
How is a Type 1 histogram reconstructed in pre-processing?
The spike at the end of the image is removed to make sure the brightness of the anatomy is accurately recorded
What are the properties of a Type 2 Histogram?
Histogram without a spike because the body part takes up the entire IR and no background radiation is recoded in the exposure field
What are the properties of a Type 3 histogram?
Histogram with two recorded spikes
- excessively bright spike on the left that represents Barium
- excessively dark spike on the right that represents background raw radiation
How is a Type 3 histogram reconstructed as part of pre-processing?
A Smin and Smax are set between the barium and raw exposure to only capture the brightness of the anatomy of interest
How are landmarks identified on a histogram?
A threshold algorithm is applied to each side of the histogram which only stores values when a certain number of pixels are reached
Once the threshold algorithm is applied what does the area between the Smin and Smax become?
The area that will be compared to the Lookup Table
What is the area between the Smin and Smax used for?
Its the Savg and is used when creating the EI
What does a grayscale curve represent?
The Log relative exposure and brightness of pixels
If a grayscale curve has a steep slope what does it represent?
The image has a short gray scale with high contrast, since its moving through the gray scale faster
If a grayscale curve has a graduated slope what does it represent?
The image has a long gray scale with low contrast, since its moving through the grayscale slower
What is gradation processing?
The selection of a VOI (volume of interest) to be sent/compared to the Lookup Table (LUT)
If a VOI on the left side of the graph is selected what will the image accentuate?
Bony anatomy at the expense of the soft tissue to the right
Why are changes to the volume of interest done?
To better highlight the type of exam you are doing which will help with the rescaling process
What is a common processing error in histogram analysis?
When selected histogram (procedural algorighm) doesn’t match the histogram acquired on the image
- Histogram mismatch
What are some secondary processing errors of histogram analysis?
- Errors with under/extreme exposure conditions
- Unique situation imaging when large prosthesis are present and included in the Smin
Why is it impotant to choose the correct anatomic region on the menu before exposing the patient?
It ensures there isn’t a histogram mismatch, which would otherwise cause issues in image brightness/contrast
What is rescaling (normalization)?
An attempt to make the image more like a conventional radiograph by simple alignments to image brightness prior to post-processing
How does the process of rescaling work?
- After initial histogram analysis, a raw iamge is created
- Software approach is completed which takes data from the Smin to Smax area, relabels them algebraically and then compares it to a LUT with pre-determined Q values
Based on the LUT table provided, at what level does Q2 output?
At a grey level of 512
Based on the LUT table provided, at what level does Qmax output?
At a grey level of 1534
