Learning Morphological Feature Perturbations for Calibrated Semi-Supervised Segmentation
Moucheng Xu, Yukun Zhou, Chen Jin, Stefano B Blumberg, Frederick Wilson, Neil Oxtoby, Marius De Groot, Daniel C. Alexander, Joseph Jacob
In this paper, we propose MisMatch, a novel consistency-driven semi-supervised segmentation framework which produces predictions that are invariant to learnt feature perturbations. MisMatch consists of an encoder and a two decoders. One decoder learns positive attention to the foreground regions of interest (RoI) on unlabelled images thereby generating dilated features of the foreground. The other decoder learns negative attention to the foreground on the same unlabelled images thereby generating eroded features of the foreground. We then apply a consistency regularisation on the paired predictions. MisMatch outperforms state-of-the-art semi-supervised methods on a CT-based pulmonary vessel segmentation task and a MRI-based brain tumour segmentation task. We also show that the effectiveness of MisMatch comes from better model calibration than its supervised learning counterpart.
Thursday 7th July
Poster Session 2.2 - onsite 11:00 - 12:00, virtual 15:20 - 16:20 (UTC+2)