Local contrastive loss with pseudo-label based self-training for semi-supervised medical image segmentation
Krishna Chaitanya, Ertunc Erdil, Neerav Karani, and Ender Konukoglu
(a) ACDC Dataset: It contains 100 short-axis MR-cine T1 3D volumes of cardiac anatomy acquired using 1.5T and 3T scanners. The expert annotations are provided for three structures: right ventricle, myocardium, and left ventricle. It was hosted as part of the MICCAI ACDC challenge 2017.