FUSSNet: Fusing Two Sources of Uncertainty for Semi-supervised Medical Image Segmentation

In recent years, various semi-supervised learning (SSL) methods have been developed to deal with the scarcity of labeled data in medical image segmentation. Especially, many of them focus on the uncertainty caused by a lack of knowledge (about the best model), i.e. epistemic uncertainty (EU). Besides EU, another type of uncertainty, aleatoric uncertainty (AU), originated from irreducible errors or noise, also commonly exists in medical imaging data. While previous SSL approaches focus on only one of them (mostly EU), this study shows that SSL segmentation models can benefit more by considering both sources of uncertainty. The proposed FUSSNet framework is featured by a joint learning scheme, which combines the EU-guided unsupervised learning and AU-guided supervised learning. We assess the method on two benchmark datasets for the segmentation of left atrium (LA) and pancreas, respectively. The experimental results show that FUSSNet outperforms the state-of-the-art semi-supervised segmentation methods by over 2% on Dice score for pancreas data and almost reaches the accuracy obtained in fully supervised setting for LA data.