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

2025-02-02

AI Summary

To address distributional bias and fairness deficits in medical image segmentation arising from demographic and clinical factors (e.g., age, sex, race, disease severity), this paper proposes a distribution-aware Mixture of Experts (dMoE) model grounded in optimal control theory. It is the first work to incorporate optimal control into fairness-aware medical image segmentation, explicitly learning and correcting multi-source distributional shifts while preserving individual segmentation accuracy and enhancing group-level fairness. Technically, dMoE integrates covariate-aware embedding, plug-and-play multi-architecture ensembling, and a dynamic gating mechanism. Evaluated on two public 2D benchmarks and one internal 3D dataset, dMoE achieves state-of-the-art performance, with an average Dice score improvement of 8.2% for underrepresented subgroups—significantly mitigating inter-subgroup segmentation disparities.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Computer Vision: Segmentation
• Search and Optimization: Distributed Search

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

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.

Problem

• Medical Image Analysis
• Bias Mitigation
• Fairness in Segmentation

Innovation

• dMoE
• Fairness in Medical Image Segmentation
• Control Theory Integration