UCL Discovery
UCL home » Library Services » Electronic resources » UCL Discovery

Learning Morphological Feature Perturbations for Calibrated Semi-Supervised Segmentation

Xu, Mou-Cheng; Zhou, Yu-Kun; Jin, Chen; Blumberg, Stefano B; Wilson, Frederick J; deGroot, Marius; Alexander, Daniel C; ... Jacob, Joseph; + view all (2022) Learning Morphological Feature Perturbations for Calibrated Semi-Supervised Segmentation. In: Proceedings of MIDL 2022. MLResearchPress (In press). Green open access

[thumbnail of 2203.10196v1.pdf]
Preview
Text
2203.10196v1.pdf - Accepted Version

Download (3MB) | Preview

Abstract

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-head decoders. One decoder learns positive attention to the foreground regions of interest (RoI) on unlabelled images thereby generating dilated features. The other decoder learns negative attention to the foreground on the same unlabelled images thereby generating eroded features. 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. In addition, we show that the effectiveness of MisMatch comes from better model calibration than its supervised learning counterpart.

Type: Proceedings paper
Title: Learning Morphological Feature Perturbations for Calibrated Semi-Supervised Segmentation
Event: 5th Conference on Medical Imaging with Deep Learning (MIDL 2022)
Open access status: An open access version is available from UCL Discovery
DOI: 10.48550/arXiv.2203.10196
Publisher version: https://proceedings.mlr.press/
Language: English
Additional information: This version is the author accepted manuscript. For information on re-use, please refer to the publisher’s terms and conditions.
Keywords: Semi-Supervised, Learning Feature Augmentation, Attention, Morphological Operations, Calibration, Consistency Regularisation, Vessel Segmentation
UCL classification: UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science
UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Med Phys and Biomedical Eng
UCL > Provost and Vice Provost Offices > UCL BEAMS
UCL
UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Computer Science
URI: https://discovery.ucl.ac.uk/id/eprint/10146545
Downloads since deposit
17Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item