The major considerations for efficacy analysis of the development of imaging product are validity and reproducibility of the results for diagnosis of disease or patient management. The validity of the diagnostic agent can be measured by sensitivity, specificity, accuracy, negative and positive predictive value comparing the diagnostic finding to a standard of truth. The sensitivity and specificity applies to different cohort of patients within a study but the two are interlinked by the disease positive rate in the study as determined by the standard of truth.

The statistical test of comparison of a new diagnostic agent with that of clinical practice can be a superiority test or a non-inferiority test. While conducting a non-inferiority test, it is important to consider the overall benefit of a new agent superior to that of existing agents in order to obtain the approval. The choice of non-inferiority margin and the effect size for the test of comparison will be very important to demonstrate overall benefit of a new agent. One way to demonstrate the overall clinical benefit would be to demonstrate superiority in one of the two diagnostic efficacy endpoint predicated upon a non-inferiority of the other endpoint to active comparator.

The selection of non-inferiority margin is complicated by the lack of placebo-controlled studies in the arena of diagnostic imaging drug development. This paper provides an approach to select a non-inferiority margin derived by a meta analysis of active comparator studies.