Statistics


TP  =  test positive in diseased subjects


FP  =  test positive in nondiseased subjects


FN  =  test negative in diseased subjects


TN  =  test negative in nondiseased subjects


T+   =  abnormal test results


T-    =  normal test results


D+   =  diseased subjects


D-   =  nondiseased subjects


STATISTICS


Incidence   =  number of diseased people per 100,000 population annually


Prevalence   =  number of existing cases per 100,000 population at a target date


Frequency   =  number of times an event occurred; often graphically represented in histograms


Mortality   =  number of deaths per 100,000 population annually


Fatality   =  number of deaths per number of diseased


Sensitivity


=  ability to detect disease


=  probability of having an abnormal test given disease


=  number of correct positive tests / number with disease


=  true positive ratio = TP / (TP + FN) = TP / D+


•  D+ column in decision matrix


◊  Independent of prevalence


Specificity


=  ability to identify absence of disease


=  probability of having a negative test given no disease


=  number of correct negative tests / number without disease


=  true negative ratio = TN / (TN + FP) = TN / D-


•  D- column in decision matrix


◊  Independent of prevalence


Accuracy


=  number of correct results in all tests


=  number of correct tests / total number of tests


=  (TP + TN) / (TP + TN + FP + FN) = (TP + TN) / total


◊  Depends much on the proportion of diseased + nondiseased subjects in studied population


◊  Not valuable for comparison of tests


Example:   same test accuracy of 90% for two tests A and B


Positive Predictive Value


=  positive test accuracy


=  likelihood that a positive test result identifies disease


=  number of correct positive tests / number of positive tests


=  TP / (TP + FP) = TP / T+


•  T+ row in decision matrix


◊  Dependent on prevalence


◊  PPV ↑ with ↑ prevalence for given sensitivity + specificity


◊  PPV ↑ with ↑ specificity for given prevalence




Negative Predictive Value


=  negative test accuracy


=  likelihood that a negative test result identifies absence of disease


=  number of correct negative tests / number of negative tests


=  TN / (TN + FN) = TN / T-


•  T- row in decision matrix


◊  Dependent on prevalence


◊  NPV ↑ with ↑ prevalence for given sensitivity + specificity


◊  NPV ↑ with ↑ sensitivity for given prevalence


False-positive Ratio


=  proportion of nondiseased patients with abnormal test result


•  D- column in decision matrix


=  FP / (FP + TN) = FP / D-


=  1 – specificity = (TN + FP – TN) / (TN + FP)


False-negative Ratio


=  proportion of diseased patients with a normal test result


•  D+ column in decision matrix


=  FN / (TP + FN) = FN / D+


=  1 – sensitivity = (TP + FN – TP) / (TP + FN)


Disease Prevalence


=  proportion of diseased subjects to total population


=  (TP + FN) / (TP + TN + FP + FN) = D+ / total



◊  Sensitivity + specificity are independent of prevalence!


◊  Affects predictive values + accuracy of a test result


Example:


Test A, C, D:   90% sensitivity + 90% specificity


Bayes Theorem


=  the predictive accuracy of any test outcome that is less than a perfect diagnostic test is influenced by


(a)  pretest likelihood of disease


(b)  criteria used to define a test result


Receiver Operating Characteristics (ROC)


=  degree of discrimination between diseased + nondiseased patients using varying diagnostic criteria instead of a single value for the TP + TN fraction


=  curvilinear graph generated by plotting TP ratio as a function of FP ratio for a number of different diagnostic criteria (ranging from definitely normal to definitely abnormal)


Y-axis:   true-positive ratio = sensitivity


X-axis:   false-positive ratio = 1 – specificity; reversing the values on the X-axis results in an identical “sensitivity-specificity curve”


Use:     variations in diagnostic criteria → reported as a continuum of responses → ranging from definitely abnormal to equivocal to definitely normal ← based on subjectivity + bias of individual radiologists


◊  A minimum of 4–5 data points of diagnostic criteria are needed!


Difficulty:   subjective evaluation of image features; subjective diagnostic interpretation; data must be ordinal (= discrete rating scale from definitely negative to definitely positive)


Interpretation:


◊  ↑ in sensitivity leads to ↓ in specificity!


◊  ↑ in specificity leads to ↓ in sensitivity!


◊  Most sensitive point is the point with the highest TP ratio


  equivalent to “overreading” by using less stringent diagnostic criteria (all findings read as abnormal)


◊  Most specific point is the point with the lowest FP ratio


  equivalent to “underreading” by using more strict diagnostic criteria (all findings read as normal)


◊  Does not consider disease prevalence in the population



◊  The ROC curve closest to the Y-axis represents the best diagnostic test

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Jun 29, 2017 | Posted by in GENERAL RADIOLOGY | Comments Off on Statistics

Full access? Get Clinical Tree

Get Clinical Tree app for offline access