Fig. 3From: Fully automated convolutional neural network-based affine algorithm improves liver registration and lesion co-localization on hepatobiliary phase T1-weighted MR imagesSmall-scale substudy (a): Registration methods and performance metrics. Binary mask (first, left to right) was considered the primary investigational method for registration and it was compared to alternate masking method, i.e., whole image registration and manual registration. Small-scale reader substudy (b): flow and performance comparisons between radiologist-performed manual registration and automated algorithm (intensity masks displayed). Three metrics: liver overlap score (Jaccard index), image correlation (mutual information), and intra-observation distanceBack to article page