Test Validation/QA Flashcards
Parameters commonly used to judge efficacy of a diagnostic test:
Sensitivity
Specificity
Accuracy
The gold standard test for carotid artery stenosis
Digital subtraction angiography
Gold standard for deep vein thrombosis
Venography
If no gold standard test, then only autopsy can be used.
T/F
False, sometimes clinical info or findings at surgery can be used
What does true positive mean?
Testing procedure demonstrates disease and disease is really present
What does true negative mean?
Test and gold standard both said there was no pathology
What does false positive mean?
Test said pathology is present, gold standard said no pathology present
False negative is when:
Test - normal
Gold standard - abnormal
How do you find the sensitivity of a test?
True Pos divided by true pos plus false neg
TP/TP+FN
Sensitivity is the ability of a test to:
Detect disease
Sensitivity measures how often a test will:
Miss real disease
Minimizing false negatives decreases sensitivity.
T/F
False, increases
How would you find specificity?
True neg divided by true neg plus false pos
TN/TN+FP
Specificity is the ability of a test to:
Identify normality or rule out pathology
Specificity is a measurement of:
How real a disease finding by that exam is
Decreasing false positives will:
Increase specificity
How do you find overall accuracy?
True pos plus true neg divided by total # of tests
TP + TN/total #
Accuracy quantifies the tests ability to:
Give correct answer
Accuracy should be left as
A fraction
What does PPV stand for?
Positive predictive value
What figures are required for PPV?
Only positive figures
How would you find PPV?
True pos divided by true pos plus false pos
TP/TP+FP
PPV tells you how meaningful a
Positive result really is
A test with a low positive predictive value cannot be used as an initial screening tool.
T/F
false, it may be useful but further testing may be recommended
PPV and NPV are represented as:
Fractions
NPV is:
Negative predictive value
How do you find NPV and what does it tell us?
True neg divided by true neg + false neg
TN/TN+FN
How meaningful a negative result really is
A test with a low negative predictive value is likely to miss disease and puts the patient in danger.
T/F
True
How would you find prevalence?
True pos plus false neg divided by total # of tests
TP+FN/Total #
Prevalence validates the test.
T/F
False, only shows the proportion of sample population that have pathology being tested for
Important qualities of an initial screening test:
High sensitivity
Low false neg rate
Low risk, low cost
Low specificity is acceptable
Qualities of a diagnostic test to determine treatment:
High specificity
Low false positive rate
Low sensitivity is acceptable if known
Example of initial screening test
ABI for arterial disease of LE, high sensitivity with little info on area/extent of disease
Duplex US or angiography can test that
Example of diagnostic test
US for gallstones, ectopic pregnancies, appendix
Tests may not detect all of pathology but surgery can be reliably planned based on pos finding
What is quality assurance?
Routine, periodic evaluation of sonography equipment and transducers
QA detects image quality problems and assures proper operation of equipment before it effects the:
Images or causes equipment to fail
Preventative maintenance is usually performed once a year by
appropriate service personnel.
T/F
False, 2-3 times
Results of QA are compared to:
Past results to identify if change is necessary to unit
Everyday maintenance of unit:
Cleaning transducers
Inspecting cables
Keeping unit dust and dirt free
File of QA documents includes:
Original equipment purchase order Equipment specs Warranty Results of QA performed Documentation of problems Follow up service report Preventative maintenance reports
Reports are needed to document the need to replace equipment but not necessary for accreditation.
T/F
False
AIUM 100 mm test object is composed of:
Series of .75 mm diameter stainless steel rods arranged in groups in transparent plastic tank filled with mixture of alcohol and water that is equal to 1.54 mm/us at room temp
The AIUM 100 mm test object is used more commonly with current technology.
T/F
False
The AIUM 100 mm test object cannot evaluate:
Gray scale or contrast resolution
Embedded in the gel of the tissue phantom are several groups of fibers which can be used to evaluate:
Axial res Lateral res Depth calibration Image uniformity Distance accuracy
Inner material of the phantom is gelatin based so in extreme temperatures the base will change its consistency but it’s still okay to use.
T/F
False
What is the dead zone ?
Distance between transducer face to area in near field that cannot be used due to reverb or transducer design
Axial res is tested by scanning rod group located:
In the middle
Placed in oblique plane
The largest separation of two closest rods seen represents axial res.
T/F
False, smallest
Lateral resolution is found by scanning the:
Same rod group as with axial res
How is system sensitivity measured?
Of the weakest echo sign visualized determined by gain setting with no TGC using the bottom rod of the group is six vertical equidistant rods
Face A of test object is used to test:
Dead zone and axial res
Face b is used to test:
Lateral res
Elevational res can be documented using:
Tissue phantom or inclined plane phantom
Limitation of using tissue phantom for slice thickness testing:
Cystic objects suspended in phantom at certain depths that cannot be changed. Elevational focus must coincide with this depth or can’t be evaluated
Inclined plane phantom uses an interface oriented at:
45 degree angle
A horizontal band is formed when testing for slice thickness. How is it measured?
The axial height equals slice thickness
Some phantoms have a series of gray scale targets that range in contrast from:
+15dB to -15dB
Doppler phantoms may contain:
A tube embedded in tissue mimicking medium connected to pump that pushes fluid through at adjustable velocities
OR
A string that moves immersed in water bath using echoes from vibrating string to imitate reflections from blood cells
Using Doppler phantoms, what can be evaluated?
Sensitivity
Sample volume accuracy
Doppler angle accuracy
Max depth of detectable Doppler signal