ABSTRACT
This chapter describes analysis of multiple-reader multiple-case (MRMC) FROC datasets. Apart from the choice of figure of merit, analyzing FROC data is similar to analyzing ROC data. The DBMH and ORH methods are applicable to any scalar figure of merit. No assumptions are made regarding independence of ratings on the same case – a sometimes misunderstood point. Analysis of a sample FROC dataset is demonstrated, including visualization of the relevant operating characteristic using RJafroc implemented functions. Suggestions are made on how to report the results of a study (the suggestions apply equally to ROC studies). Single fixed factor analysis is described, followed by a newly developed crossed-treatment analysis, applicable when one has two treatment factors and their levels are crossed. Sample size estimation for FROC studies is also not fundamentally different from that described for ROC studies. Using the RSM fitting method described in the previous chapter, NH values for the three RSM parameters are derived. These allow relating a chosen ROC effect size to the equivalent AFROC effect size. The latter is usually larger, hence the increased power.
