Skip to main content
Fig. 4 | Radiation Oncology

Fig. 4

From: 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 network

Fig. 4

The best (a) and the worse (b) segmentation result shown for total bone marrow, bladder, bowel cavity, small bowel, large bowel, and all bowel structures. Test cases were chosen based on the minimum and maximum average Dice similarity coefficient (DSC). All presented structures were in good agreement with the ground truth (GT, expert delineation) for the best test case, resulting in a mean DSC of 0.94. For the worst case, a mean DSC of 0.8 was found, mainly due to inaccurate segmentations of the small and large bowel structures, indicated by the yellow arrows. Parts of the small bowel were wrongly classified as parts of the large bowel and vice versa

Back to article page

Radiation Oncology

ISSN: 1748-717X

Contact us