Disentangling Human Error from the Ground Truth in Segmentation of Medical Images Le Zhang

Recent years have seen increasing use of supervised learning methods for segmentation tasks. However, the predictive performance of these algorithms depends on the quality of labels.