Disentangled Representations for Domain-generalized Cardiac Segmentation

TL;DR

This paper introduces two novel data augmentation techniques to enhance the domain generalization and adaptation of cardiac segmentation models, aiming to improve performance on unseen medical imaging data.

Contribution

The paper proposes 'Resolution Augmentation' and 'Factor-based Augmentation' methods that generate diverse training data to improve model robustness across different domains.

Findings

Enhanced segmentation accuracy on unseen domains.

Improved model robustness through data diversity.

Effective domain adaptation demonstrated in experiments.

Abstract

Robust cardiac image segmentation is still an open challenge due to the inability of the existing methods to achieve satisfactory performance on unseen data of different domains. Since the acquisition and annotation of medical data are costly and time-consuming, recent work focuses on domain adaptation and generalization to bridge the gap between data from different populations and scanners. In this paper, we propose two data augmentation methods that focus on improving the domain adaptation and generalization abilities of state-to-the-art cardiac segmentation models. In particular, our "Resolution Augmentation" method generates more diverse data by rescaling images to different resolutions within a range spanning different scanner protocols. Subsequently, our "Factor-based Augmentation" method generates more diverse data by projecting the original samples onto disentangled latent spaces, and combining the learned anatomy and modality factors from different domains. Our extensive experiments demonstrate the importance of efficient adaptation between seen and unseen domains, as well as model generalization ability, to robust cardiac image segmentation.