Lecture 2 Flashcards
(16 cards)
Explain the relationship between scatter, noise and contrast
Scatter is caused by photons that are scattered in angles
Ie. moving away from the pt and across the pt
These scattered photons have a lower energy than primary photons, so they combine with lighter signals and reduce the ability of high energy beta signals to produce a good image
Therefore, increase scattered photons = decrease contrast because low energy photons contribute to the image but don’t carry useful info and thus, will become scatter
Noise is differnt, as it regards only a specific region of the image.
Increase in photons in a region = decrease in noise.
The amount of photon fluctuation (ie. SD or noise value) is related to average photon concentration, or exposure level.
Increase photon concentration (SD) = decrease in noise
SD= % of average photon concentration
Define the methodologies of evaluation of noise
Can be assessed quantitatively, and qualitatively
Quality = obeserver visual rating
Quan = ROI selected to identify SI, SD for regions
Explain noise evaluation
Differences in attenuation can cause differences in quantity of photons reaching the detector at differnt parts of image. So to find SD of one region we select that region and measure noise in terms of signal intensity to get the SD (noise level).
SD is the NPS (absolute noise) And tells us the noise characteristics of the image
Increase SD = increase in noise
SD is just the fluctuation
SNR is used to find impact of that noise on image quality
Ratio calculated as the SI of ROI/ SD
Good image quality has SNR > 1
High SNR = increase image quality (but might have insufficient overall contrast)
Explain contrast evaluation
Conducted in relevance to noise, as SNR provides measure of the effect of noise on contrast at ROI
So measure SNR at region at differnt regions of image (similar attenuation regions). So CNR characterises quality of image, and tells us degradation of contrast due to noise. But doesn’t asses the imaging system itself
So to Calculate: 2 ROI and find their signal intensities relative to background noise. SD is the background noise
(Sa-Sb)/SD
Explain the European quality guideline for image creiteria
There are requirements for it to be a quality image
Ie. visualisation, reproduction, visually sharp reproduction, important image details
Explain Image criteria scaling
Observer views and rates image on basis of fulfilment of criteria
Ie. the European criteria
Depending on how fulfilled these criteria are in the observers opinion determine the image quality
What are the advantages of ICS?
Corresponds to a clinically acceptable level of image quality
Both structures marked as important and the level of reproduction can be used consistently (each clinic has its own criteria)
What are the limitations of ICS?
Soft transition from “not fulfilled” to fulfilled
Very subjective
Explain Visual Grading Analysis
There’s 2 forms of this, absolute and relative
Absolute has 4 criteria
- Poor image quality
- Restricted image quality
- Sufficient image q
- Good IQ
- Excellent IQ
Relative is when you compare a test image to a standard. Can be used to justify dose as well
- Test image is inferior to ref image
- Somewhat inf to ref image
- Equal
- Somwhat superior
- Superior to ref image
What are the advantages of relative VGA?
Used to compare different protocols
Ie. low exposure protocol compared to high exp protocol can determine whether low exp protocol still gives us good IQ in comparison to high exp (thus, low dose will be fine)
Helps investigate dose reduction possibilities
What are the limitations of relative VGA?
Subjective
Bias and inter observer variability suspected
Explain what’s observer performance studies
Observers interpret a set of normal and abnormal images
Performance in correctly classifying normal and abnormal is calculated using differnt observer performance analysis like ROC and JAFROC
Explain ROC analysis
Form of observer performance studies
Provides measure of sensitivity (number of cases correctly identified with pathology) and specificity (number of cases identified normal correctly)
We use AUC to characterise overall global performance
What does the AUC tells us?
High AUC = more detections of pathologies, so good IQ
For this, we are looking for reproduction of structures rather than contrast, noise, resolution.
0.90-1 = excellent A >0.80-0.90= good B >0.70-0.80 = fair C > 0.60-0.70=poor D >0.50-0.60=fail E
Explain VGCA (visual grading characteristic analysis)
Extension of ICS where observer uses a multi-step rating scale to see if IQ criteria has been fulfilled. Provide criteria
Good for comparing 2 imagejng protocols.
Was introduced to distinguish internal conflict of observer without having to be for certain eg. Rating scale
- Confident that the criteria is not fulfilled
- Somewhat confident that is not fulfilled
- Indecisive whether fulfilled
- Somewhat confident that it is fulfilled
- Confident that criteria is fulfilled
Explain the steps in VGC analysis
1.
