Radiology and Probability

Like every other test in medicine, radiology tests are ones of probability. This is entirely evident in the way radiological impressions are made: "chest X-ray findings are consistent with pneumonia," "spiculated breast mass concerning for malignancy." Few things in medicine are diagnostic and most tests are functions of the prevalence of the diagnosis in question along with the sensitivity and specificity of the test. The problem is that many people, especially the general public, think of radiologic tests as perfectly precise and all-or-nothing. False positives (a radiologist seeing a finding when one is not there) and false negatives (a radiologist missing a diagnosis) are inherent in tests; very few tests are such that mistakes are impossible. Indeed, a radiologist goes through many years of training and a lifetime of practice to hone those skills and minimize false positives and negatives. Yet because patients sometimes expect 100% certain, black-or-white (pun not intended) diagnoses, lawsuits against radiologists are a major problem.

I am fascinated by this because it colors radiology reports. They may hedge when they are leaning towards a diagnosis but know they may be calling a false positive. They qualify their statements when false negatives are possible ("no evidence of malignancy seen on suboptimal study"). But I think the bottom line is this: radiological findings must be correlated with the clinical picture because both change the probabilities of various diagnoses. And radiologists, like any other doctor, are not perfect.