FairREAD: Re-fusing Demographic Attributes after Disentanglement for Fair Medical Image Classification

2024-12-20

arXiv.org

AI Summary

In medical image classification, removing sensitive attributes (e.g., demographic features) often erodes clinically relevant information, hindering simultaneous improvement of fairness and diagnostic performance. To address this, we propose a “decouple-then-controllably-refuse” paradigm: first, orthogonal constraints and adversarial training disentangle bias-laden and clinically informative components within sensitive attributes; second, a controllable feature re-fusion mechanism selectively preserves the latter; third, subgroup-specific adaptive decision thresholds are introduced to jointly optimize fairness and diagnostic accuracy. Evaluated on a large-scale chest X-ray dataset, our method reduces equalized odds disparity by over 60% while maintaining baseline diagnostic accuracy. This work establishes the first framework for separable modeling and selective utilization of bias versus clinical signals embedded in sensitive attributes, offering a novel direction for fairness-aware medical AI.

Technology Category

• Computer Vision: Bias, Fairness & Privacy
• Machine Learning: Ethics, Bias, and Fairness
• Philosophy and Ethics of AI: Bias, Fairness & Equity

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
• Security and Privacy: Data transparency and provenance

Abstract

Recent advancements in deep learning have shown transformative potential in medical imaging, yet concerns about fairness persist due to performance disparities across demographic subgroups. Existing methods aim to address these biases by mitigating sensitive attributes in image data; however, these attributes often carry clinically relevant information, and their removal can compromise model performance-a highly undesirable outcome. To address this challenge, we propose Fair Re-fusion After Disentanglement (FairREAD), a novel, simple, and efficient framework that mitigates unfairness by re-integrating sensitive demographic attributes into fair image representations. FairREAD employs orthogonality constraints and adversarial training to disentangle demographic information while using a controlled re-fusion mechanism to preserve clinically relevant details. Additionally, subgroup-specific threshold adjustments ensure equitable performance across demographic groups. Comprehensive evaluations on a large-scale clinical X-ray dataset demonstrate that FairREAD significantly reduces unfairness metrics while maintaining diagnostic accuracy, establishing a new benchmark for fairness and performance in medical image classification.

Problem

• Mitigates unfairness in medical image classification across demographics
• Preserves clinically relevant information while removing biases
• Ensures equitable performance without compromising diagnostic accuracy

Innovation

• Disentangles demographic info via orthogonality constraints
• Re-fuses attributes to preserve clinical relevance
• Adjusts thresholds for equitable subgroup performance