Neuron Incidence Redistribution for Fairness in Medical Image Classification

TL;DR

This paper introduces Neuron Incidence Redistribution (NIR), a regularization method that reduces demographic disparities in medical image classification by redistributing neural activation patterns without needing demographic labels.

Contribution

The paper proposes NIR, a novel regularization technique that improves fairness in medical AI models by balancing neuron activations, without requiring demographic annotations during training.

Findings

NIR significantly reduces demographic performance disparities in medical image classification.

NIR improves fairness metrics on HAM10000 and Harvard OCT-RNFL datasets.

NIR achieves these improvements with minimal impact on overall accuracy.

Abstract

Deep learning models for medical image classification are susceptible to subgroup performance disparities across demographic attributes such as age, gender, and race. We identify a latent representational mechanism underlying these disparities: in transfer-learned models, the dominant penultimate-layer activation channel under positive predictions is co-activated by both disease-positive samples and privileged demographic groups (male, older patients), producing over-diagnosis; conversely, the dominant channel under negative predictions is co-activated by disadvantaged groups (female, younger patients), producing systematic under-diagnosis. To address this, we propose Neuron Incidence Redistribution (NIR), a lightweight regularization method that penalizes the variance of predicted-probability-weighted mean activations across penultimate-layer neurons, requiring no demographic labels at training time. On HAM10000, TPR disparity drops from 10.81% to 0.93% across age groups and from 12.04% to 0.74% across gender, with a marginal AUC improvement of 0.51 points. On Harvard OCT-RNFL, NIR reduces FPR disparity for race (from 15.68% to 10.66%) and age (from 12.69% to 1.80%), demonstrating that distributing latent disease evidence across the full penultimate layer is a principled and effective strategy for improving demographic fairness in medical AI.