Distribution-aware Fairness Learning in Medical Image Segmentation From A Control-Theoretic Perspective

TL;DR

This paper introduces a distribution-aware mixture of experts (dMoE) model inspired by control theory to improve fairness in medical image segmentation by addressing demographic and clinical biases, achieving state-of-the-art results.

Contribution

The paper proposes a novel control-theoretic dMoE framework for fairness-aware medical image segmentation, applicable across multiple architectures and datasets.

Findings

dMoE achieves state-of-the-art performance on benchmark datasets.

Incorporating demographic and clinical factors reduces bias in segmentation.

The approach demonstrates broad applicability across diverse medical imaging tasks.

Abstract

Ensuring fairness in medical image segmentation is critical due to biases in imbalanced clinical data acquisition caused by demographic attributes (e.g., age, sex, race) and clinical factors (e.g., disease severity). To address these challenges, we introduce Distribution-aware Mixture of Experts (dMoE), inspired by optimal control theory. We provide a comprehensive analysis of its underlying mechanisms and clarify dMoE's role in adapting to heterogeneous distributions in medical image segmentation. Furthermore, we integrate dMoE into multiple network architectures, demonstrating its broad applicability across diverse medical image analysis tasks. By incorporating demographic and clinical factors, dMoE achieves state-of-the-art performance on two 2D benchmark datasets and a 3D in-house dataset. Our results highlight the effectiveness of dMoE in mitigating biases from imbalanced distributions, offering a promising approach to bridging control theory and medical image segmentation within fairness learning paradigms. The source code will be made available. The source code is available at https://github.com/tvseg/dMoE.