UNIT 3: Post-Processing Flashcards
Is the Image receptor response to exposure linear or not?
Linear response
Diagnostic x-ray
Acceptable level of noise with the lowest possible patient dose
T or F: As CNR decreases, spatial resolution decreases
True
Is higher or lower CNR better?
Higher
Is higher or lower SNR better?
Higher
Spatial domain
Sorts the information by location on the Detector
Intensity domain
Sorts the information by pixel value (gray shade)
Frequency domain
Sorts the information by size of structure
T or F: small objects have a low spatial frequency
False, they have a high spatial frequency. Large objects have a low spatial frequency
Kernels are also called
Masks or Filters
Image to Image transformation: Spatial Resolution
Input image —> Kernel —> Output image
Image to Image transformation: Frequency Domain
Input image —> Fourier transform —> Frequency filter —> Reverse Fourier transform —> Output image
Operator adjustments made to a processed image
-Window level
-Image Inversion
-Edge enhancement
-Image Re-orienting(flip or rotate)
-Image Zoom
-Annotation
Multifrequency processing
-The computer places contrast and noise in a separate layer, which can be removed from the final image
-Can separate the image into frequency or structural size components called layers or band and each layer can be processed separately from other layers
Steps of Post-Processing
- Default Gradation processing (application of the LUT)
- Detail processing
- Formatting for display
- Image Displayed
- Operator adjustments
What is gradation processing/gradient processing? What does gradient mean?
-Grayscale processing for contrast of an image via the Look Up Table. It also refers to changes in the contrast scale of the image. Also affects the slope of the contrast meaning does the image need to be more short/high scale contrast with a steeper slope or long scale contrast.
-Gradient: describes the rate of change in a scale or how steep the slope is
What is the purpose of the LUT?
LUT is a set of output values and the mathematical calculations that the input values are subjected to in order to produce image contrast. A table of numbers used to perform gradient functions on the image to change the grayscale latitude or slope of the contrast curve
T or F: image brightness or contrast are primarily determined by the LUT rather than the exposure to the IR
True
Data clipping
• Narrowing of the range of values
• This occurs if the image is overly adjusted, resulting in some data loss
How might data clipping negatively affect image quality?
Occurs when the technologist windows the image too much at the QC station prior to sending the image to PACS. Data clipping limits the radiologists ability to window the image and can affect diagnosis because it can make the contrast very low which would make it harder for the radiologist to interpret. Techs should avoid sending images to PACS that have been windowed at the QC station
What is CNR (Contrast to Noise Ratio)?
The ratio of the difference of signal intensities of 2 regions of interest to the imaging noise and is dependent upon both digital image contrast and signal to noise ratio (SNR). In other words, it is irrelevant contrast (difference between densities) that exists in the image. So the measure of the contrast between tissue of interest and background
How is CNR measured?
Measured using ROI (Region of Interest) which is a quantitative measurement of the pixel values within a selected region
Detail processing is performed to
Correct for image noise and enhance certain structures
In which 2 domains might detail processing occur?
-Spatial domain
-Frequency domain
When detail processing is performed in the spatial domain, what type of algorithm is used?
It is accomplished using kernels using linear processing techniques. A linear filtering technique occurs when the output image is created using a weighted sum of the input pixels
What are the advantages and disadvantages of high pass filtering and what is it?
-High pass filtering: allows the high frequency components of the image to remain and filter out the lower frequencies. This enhances the detail structures and edges of tissues, aka edge enhancement
-Advantages: Image detail is increased
-Disadvantages: More noise on the image
What are the advantages and disadvantages of low pass filtering and what is it?
-Low pass filtering: allows low frequency components of the image to remain and filter out the higher frequencies. This removes fine detail in the image so the image appears more blurry and smooth, aka smoothing
-Advantages: Less noise on the image
-Disadvantages: Image detail is decreased
Digital subtraction is also referred to as
unsharp mask
Explain how an unsharp mask improves image quality. What is the disadvantage?
• It is going to enhance detail in certain clinical procedures especially for viewing blood vessels or in chest radiography
• Disadvantage:
Define Periodic Noise and Provide one example of periodic noise
• Periodic noise(nonrandom noise): small artifacts of consistent size and pattern.
• Ex: electronic noise, TFT’s
What types of algorithms may be used to correct Periodic noise
Using the smoothing algorithm in the frequency domain can correct it
Define Random Noise and Provide one example of random noise
• Random noise: artifacts of varying sizes that occur in irregular patterns.
• Ex: quantum mottle
What types of algorithms may be used to correct Random noise
MUST be fixed in the spatial domain using kernels
2 ways to achieve dual energy subtraction
- Use a double exposure technique- one exposure taken at high kVp and one at Low KVp (GE). This requires the patient hold their breath for both exposures. The system takes about 1/5 sec (200 milliseconds) to switch between exposures.
- A filter is placed between two or more image receptor plates. The first plate records the low energy beam, the beam hardens through the filter and the second plate records the high energy beam. The data from the two plates are combined by the computer.
What is dual energy subtraction?
The production of bone only and soft tissue only images by using 2 different exposures at different energy levels
What is Image stitching and what is it commonly used for?
-Image stitching is taking multiple images and putting them together to appear as if they were created just using one exposure
-Scoliosis and bone length studies
What is Virtual Grid software? What is the advantage of it?
-Software that produces images appearing to have been taken using a grid when actually no grid was used
-Advantage: provide additional image post-processing capabilities that help reduce patient exposure while improving image quality
How is Digital Subtractions Angiography performed?
A non-contrast enhanced image is used as a digital mask over a contrast enhanced image, then the non-contrast enhanced image is removed or subtracted, leaving only the contrast enhanced image
What is Digital Subtraction Angiography used for?
To visualize and increase contrast in intravenous contrast filled blood vessels. Used often in fluoroscopy to examine the flow of blood through blood vessels to look for pathologies such as stenosis, occlusion or blockage, etc
Brightness
measurement of the luminance of an area in a radiographic image displayed on a monitor
Contrast
the visible difference between any two selected areas of brightness levels within the displayed radiographic image. It is determined primarily by the processing algorithm (mathematical codes used by the software to provide the desired image appearance)
Grayscale
refers to the number of brightness levels (or gray shades) visible on an image and is linked to the bit depth of the system.
Long Scale
the term used when slight differences between gray shades are present (low contrast) but the total number of gray shades is great
Short Scale
the term used when considerable or major differences between gray shades are present (high contrast) but the total number of gray shades is small
Windowing
-used when describing manual (by the operator) adjustments to the brightness and contrast of the image displayed at the monitor screen with the controls at the console as the image is viewed
-There are two windowing parameters the operator can manipulate after initial default grayscale processing called Window Width and Window Level
Window Width (WW)
changes the contrast latitude, scale or the number of gray shades represented in the image on the computer monitor. Graphically, this may also be referred to as the slope of the contrast curve. Changes in window width does not change the image brightness
Window Level
Changes the center gray level of the window width or the average brightness (depends on the vendor). This parameter changes image brightness but does not change the scale or the number of gray shades visible
T or F: If you window an image, you should reset to the default state prior to sending to PACS
True, because windowing can cause data to be clipped out and results in data loss
Gradation processing refers to changes in the
Contrast scale in an image
Graph of high/short scale contrast and medium/short scale contrast curves
-High/short scale contrast: curve is steeper
-Medium/short scale contrast: curve is flatter
Dynamic range
range of different gray levels made available by a computer system including its installed software.
If the dynamic range or bit depth of a digital processing system is too limited, it is possible for ______ __________ to occur when adjustments to contrast or brightness are made
Data clipping
T or F: it is important for radiographers to window an image before saving it in PACS
False, avoid windowing
Two main types of detail processing are
edge enhancement and smoothing
Smoothing
removes the fine edges of structures, removes noise and removes detail
For smoothing or noise reduction, using ________ in the spatial domain is better suited for removing random noise, such as quantum mottle from the image
Kernels
low pass filtering is better suited for removing
Periodic noise such as electronic noise
Edge Enhancement
actuates the line detalis or structures, enhances detail and increases noise
In the spatial domain detail processing is accomplished using _________ using linear processing techniques.
kernels (AKA masks or filters)
two main types of linear kernel processing
Correlation and Convolution
Digital subtraction process
The process consists of two distinct steps.
-First, a blurred copy of the original image is produced. A common form of digital processing that can be used to produce a blurred image is to replace each pixel value with the average of the pixel values in its neighborhood (i.e., kernel processing). This produces a “blurred mask” image. The blurring removes some or all of the detail components from the image. We now have two images.
The original image contains the general large-area contrast background plus some detail. The blurred mask image contains only the large-contrast background
-Second, the computer subtracts the blurred mask image (actually some fraction of it) from the original image. This process reduces the large-area contrast background in relation to the contrast of the detail. The result is that the contrast and visibility of the detail (small objects and structures in enhanced
Unsharp mask aka digital subtraction
Computer subtracts the blurred mask image from the original image. This process reduces the large-area contrast background in relation to the contrast of the detail. The result is that the contrast and visibility of the detail (small objects and structures) in enhanced
Frequency processing, such as low pass filtering is better suited for removing _________ noise, such as electronic noise
Periodic