Statistics refresher for molecular imaging technologists, part 1: Testing the test

Abstract

Molecular imaging technologists face statistics every day because they perform diagnostic imaging on patients who have been selected on the basis of how well the imaging test performs and because that performance can be influenced by a variety of technical factors. Choosing the test for a patient requires an understanding of the patient's pretest likelihood of disease, the performance of the test, and other clinical factors that may affect the results. Terms such as sensitivity, specificity, accuracy, and predictive value are used to describe how a diagnostic test performs or how it compares with other diagnostic tests. Although nuclear medicine and PET technologists study these concepts in training programs, applying this learning to daily patient care can be daunting given that new tracers and technologies are continuously entering clinical practice. Advances in research continue to challenge diagnostic paradigms with new patient populations, increasingly sophisticated technology, and the advent of large databases with which to study population outcomes. This article-part 1 of a 2-part series-is a refresher on the basic clinical statistics that are useful in understanding how diagnostic testing is optimally applied in patient care.