aka the JAFROC website


Raw FROC: consider, figure below, laying out the marks in two rows, one for NLs - lesion level "false positives" denoted by the red circles below, and one for LLs - lesion level "true positives" denoted by the green circles below. The quotes are used to denote that these ROC-specific terms should never be used in the FROC context. The marks are ordered with the confidence level (z) increasing to the right. temp1Starting on the extreme right hand side, from positive infinity, one moves an imaginary cutoff slowly to the left. The first mark hit may be a LL (green circle) and just crossing it will yield an FROC operating point at (0,1/L) where L is the total number of lesion s in the dataset. On keeping moving the threshold zeta, as one crosses each LL the operating point moves up by 1/L and as one crosses each NL the operating point moves to the right by 1/I, where I is the total number of images. This one gets a saw tooth curve, called the raw FROC curve. Why is it useful? It allows visual determination of whether the observer is fully using the rating scale, see two examples below. Ideally the raw FROC curve should end in a plateau. [The number of points on the raw FROC curve equals the total number of marks. If binning is employed the number of FROC points will be smaller than the number of marks.]tempo2 djklajdslka

Why is it not useful? See below. Over reliance on this curve is one of the reasons for CAD's dismal performance in breast mass detection.


Creator: Dev P. Chakraborty, Ph. D. | ©2005, ©2017, ©2018, ©2019