Many of the investigations that are now carried out during imaging systems' evaluations and comparisons include interpretation (reading) studies of normal and abnormal cases that are evaluated by receiver operating characteristics (ROC) or derivative thereof. Unfortunately, if one takes into account case selection representation and reader variability most studies would require a prohibitively large sample size to produce adequate statistical power for a statistically significant discrimination between reading modes in a manner which is both clinically relevant and generalizable. Approaches to case selection protocols for observer performance studies in this field have been largely arbitrary. We propose to investigate the possibility that the implementation of specific case selection protocols could allow one to focus on limited clinical questions which can be conclusively answered with an observer performance study requiring a much smaller sample size; hence, making such studies much more feasible and reasonable to perform. In addition, in conjunction with data on the proportion of such cases in the clinical environment (general ensemble of images), generalization of the results and or inference on the impact on clinical practices could be done.
Showing the most recent 10 out of 16 publications
