Semi-supervised Pathology Segmentation with Disentangled Representations

TL;DR

This paper introduces APD-Net, a semi-supervised model that disentangles anatomy, modality, and pathology representations to improve pathology segmentation with limited labeled data.

Contribution

The novel contribution is the joint learning of anatomy, modality, and pathology disentanglement in a semi-supervised framework for improved segmentation.

Findings

APD-Net outperforms existing methods on cardiac infarction datasets.

The model maintains performance with varying amounts of supervision.

It effectively segments pathology with few annotations.

Abstract

Automated pathology segmentation remains a valuable diagnostic tool in clinical practice. However, collecting training data is challenging. Semi-supervised approaches by combining labelled and unlabelled data can offer a solution to data scarcity. An approach to semi-supervised learning relies on reconstruction objectives (as self-supervision objectives) that learns in a joint fashion suitable representations for the task. Here, we propose Anatomy-Pathology Disentanglement Network (APD-Net), a pathology segmentation model that attempts to learn jointly for the first time: disentanglement of anatomy, modality, and pathology. The model is trained in a semi-supervised fashion with new reconstruction losses directly aiming to improve pathology segmentation with limited annotations. In addition, a joint optimization strategy is proposed to fully take advantage of the available annotations. We evaluate our methods with two private cardiac infarction segmentation datasets with LGE-MRI scans. APD-Net can perform pathology segmentation with few annotations, maintain performance with different amounts of supervision, and outperform related deep learning methods.