Automated segmentation methods for liver analysis in oncology applications
László Ruskó · 2014
This measure demonstrates only the under-segmentation.It is equal to 100% if there is no under-segmentation, and 0% if the intersection of and is empty.This measure demonstrates only the over-segmentation.It is equal to 0% if there is no over-segmentation, and arbitrary large number depending of the size of the oversegmentation.Note that neither nor can characterize segmentation accuracy alone, both metrics shall be always considered. Dice Similarity Coefficient (%):