Figure 3
From: Ranked retrieval of segmented nuclei for objective assessment of cancer gene repositioning
(a) ROC curves for each leave-one-out experiment trial, with the ground truth taken from all three reviewers (b) ROC performance as a function of the number of reviewers (c) Sorted values of the posterior probability values for each dataset, scaled to the domain [0,1]. The area under these curves is a representation of the yield of well segmented nuclei from the dataset. (d) The effect of training set size on the regression performance, reported for 500 repetitions.