Review of the new methodology of analyzing receiver operating characteristic curves

The Dorfman and Alf algorithm based on the classic model of signal detection theory is most frequently used for the receiver operating characteristic (ROC) analysis. However, several shortcomings of this method were pointed out. On these problems, in recent years, several authors have proposed the new analysis methods of the ROC curve. This paper reviews these new analysis methods, and discusses them. First, we review the general regression methodology for the ROC analysis, proposed by Tosteson and Begg. Secondly, we review the analysis of variance (ANOVA) method, proposed by Dorfman et al; in this method, the jackknife method was applied to ANOVA for analyzing ROC data. Further, we give the illustrative experimental design, and discuss the application of these methods to our illustrative experiment. We indicate that we can make use of these two methods for the evaluation of diagnostic performance in our illustrative experiment.