Semantic segmentation of computed tomography for radiotherapy with deep learning: Compensating insufficient annotation quality using contour augmentation

摘要

In radiotherapy treatment planning, manual annotation of organs-at-risk and target volumes is a difficult andtime-consuming task, prone to intra and inter-observer variabilities. Deep learning networks (DLNs) are gain-ing worldwide attention to automate such annotative tasks because of their ability to capture data hierarchy.However, for better performance DLNs require large number of data samples whereas annotated medical data isscarce. To remedy this, data augmentation is used to increase the training data for DLNs that enables robustlearning by incorporating spatial/translational invariance into the training phase. Importantly, performance ofDLNs is highly dependent on the ground truth (GT) quality: if manual annotation is not accurate enough, thenetwork cannot learn better than the annotated example. This highlights the need to compensate for possiblyinsufficient GT quality using augmentation, i.e., by providing more GTs per image, in order to improve perfor-mance of DLNs. In this work, small random alterations were applied to GT and each altered GT was consideredas an additional annotation. Contour augmentation was used to train a dilated U-Net in multiple GTs perimage setting, which was tested on a pelvic CT dataset acquired from 67 patients to segment bladder and rectumin a multi-class segmentation setting. By using contour augmentation (coupled with data augmentation), thenetwork learnt better than with data augmentation only, as it was able to correct slightly offset contours inGT. The segmentation results produced were quantified using spatial overlap, distance-based and probabilisticmeasures. The Dice score for bladder and rectum are reported as 0:88-0:19 and 0:89-0:04, whereas the averagesymmetric surface distance are 0:22 - 0:09 mm and 0:09 - 0:05 mm, respectively.