Fig. 3From: Pelvic U-Net: multi-label semantic segmentation of pelvic organs at risk for radiation therapy anal cancer patients using a deeply supervised shuffle attention convolutional neural networkBox and Whisker plot showing the result of the quantitative segmentation evaluation for the test dataset (\(n = 15\)). (a) Dice similarity coefficient (DSC), (b) the 95th percentile of the Hausdorff distance (HD95) and (c) the mean surface distance (MSD), were computed as the mean value over the test data. Overall, all segmented structures led to high DSC values, while some outliers were observed for small and large bowel structuresBack to article page