Auto-segmentations by convolutional neural network in cervical and anorectal cancer with clinical structure sets as the ground truth

Abstract

Background: It is time-consuming for oncologists to delineate volumes for radiotherapy treatment in computer tomography (CT) images. Automatic delineation based on image processing exists, but with varied accuracy and moderate time savings. Using convolutional neural network (CNN), delineations of volumes are faster and more accurate. We have used CTs with the annotated structure sets to train and evaluate a CNN. Material and methods: The CNN is a standard segmentation network modified to minimize memory usage. We used CTs and structure sets from 75 cervical cancers and 191 anorectal cancers receiving radiation therapy at Skåne University Hospital 2014-2018. Five structures were investigated: left/right femoral heads, bladder, bowel bag, and clinical target volume of lymph nodes (CTVNs). Dice score and mean surface distance (MSD) (mm) evaluated accuracy, and one oncologist qualitatively evaluated autosegmentations. Results: Median Dice/MSD scores for anorectal cancer: 0.91-0.92/1.93-1.86 femoral heads, 0.94/2.07 bladder, and 0.83/6.80 bowel bag. Median Dice scores for cervical cancer were 0.93-0.94/1.42-1.49 femoral heads, 0.84/3.51 bladder, 0.88/5.80 bowel bag, and 0.82/3.89 CTVNs. With qualitative evaluation, performance on femoral heads and bladder auto-segmentations was mostly excellent, but CTVN autosegmentations were not acceptable to a larger extent. Discussion: It is possible to train a CNN with high overlap using structure sets as ground truth. Manually delineated pelvic volumes from structure sets do not always strictly follow volume boundaries and are sometimes inaccurately defined, which leads to similar inaccuracies in the CNN output. More data that is consistently annotated is needed to achieve higher CNN accuracy and to enable future clinical implementation.

Machine‐assisted interpolation algorithm for semi‐automated segmentation of highly deformable organs

“…Compared to the state-of -the-art automatic segmentation methods involving the stomach 23,25,[32][33][34][35][36] (reported DSC 0.81-0.90) and bowel 25,[37][38][39][40][41] (reported DSC 0.78-0.89), MAI shows higher segmentation accuracy. Because auto-segmentation results are sensitive to the composition of training and evaluation datasets, we also compared MAI to a fully automated Dense-UNet segmentation that was trained and evaluated on our own dataset.…”

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

“…Although direct comparisons with other studies should be undertaken with caution, we believe that the results obtained in this work compare favorably with previous studies of deep learning-based OAR segmentations in the pelvic region. In a recent study, Sartor et al [ 25 ] trained a 3D U-Net-like architecture using CT volumes of 191 anorectal cancer patients. The lower mean DSC values in that study compared to our study (e.g., bowel cavity 0.82 vs. 0.95) could be attributed either to differences in model performance or differences in the dataset.…”

“…Even though many researchers have incorporated neural network-based segmentation methods for RT purposes, only a few models for automatic segmentation of pelvic OAR and target structures have been proposed [ 18 , 23 – 25 ]. Notably, automated deep learning-based segmentation methods have rarely been applied to complex pelvic OAR structures, like small and large bowel, and resulted in relatively unsatisfactory segmentation metrics [ 23 – 25 ].…”

Unsupervised Domain Adaptation for Small Bowel Segmentation using Disentangled Representation

“…Over the years, there have been attempts to develop automatic methods for small bowel segmentation, especially using deep learning. The small bowel was included in segmenting multiple organs-at-risk for radiotherapy treatment planning of affected tissues, such as pancreatic and cervical cancers, in CT scans [8,9,13]. Although the results obtained for the small bowel are reasonable, some of their data included only the part of the small bowel that is closest to the target area, which needed to be dose-evaluated [8,9].…”

“…Although the results obtained for the small bowel are reasonable, some of their data included only the part of the small bowel that is closest to the target area, which needed to be dose-evaluated [8,9]. In [13], the rough bowel location was detected instead of performing pixel-accurate small bowel segmentation. There have been only a few previous works dedicated solely to automatic small bowel segmentation [12,14,21].…”