Adversarial Consistency for Single Domain Generalization in Medical Image Segmentation

TL;DR

This paper introduces an adversarial domain generalization approach for medical image segmentation that trains on a single domain and synthesizes diverse domains to improve generalization to unseen data.

Contribution

It proposes a novel adversarial domain synthesizer and a mutual information regularizer to enable single-domain training for effective organ segmentation across unseen modalities.

Findings

Effective segmentation on unseen modalities and scanner settings.

Synthesizes diverse domains to cover plausible distribution space.

Improves generalization without multi-domain training data.

Abstract

An organ segmentation method that can generalize to unseen contrasts and scanner settings can significantly reduce the need for retraining of deep learning models. Domain Generalization (DG) aims to achieve this goal. However, most DG methods for segmentation require training data from multiple domains during training. We propose a novel adversarial domain generalization method for organ segmentation trained on data from a \emph{single} domain. We synthesize the new domains via learning an adversarial domain synthesizer (ADS) and presume that the synthetic domains cover a large enough area of plausible distributions so that unseen domains can be interpolated from synthetic domains. We propose a mutual information regularizer to enforce the semantic consistency between images from the synthetic domains, which can be estimated by patch-level contrastive learning. We evaluate our method for various organ segmentation for unseen modalities, scanning protocols, and scanner sites.