Evaluating the Fairness of Neural Collapse in Medical Image Classification

TL;DR

This paper investigates how Neural Collapse, a phenomenon in deep learning, affects fairness in medical image classification, revealing that bias influences NC dynamics and can reduce subgroup performance.

Contribution

It explores the relationship between Neural Collapse and bias in medical imaging, showing how biased training impacts NC configurations and subgroup fairness.

Findings

Biased training causes different NC configurations across subgroups.

NC converges to a final solution by memorizing all data samples.

Bias in training leads to a significant drop in F1 scores across subgroups.

Abstract

Deep learning has achieved impressive performance across various medical imaging tasks. However, its inherent bias against specific groups hinders its clinical applicability in equitable healthcare systems. A recently discovered phenomenon, Neural Collapse (NC), has shown potential in improving the generalization of state-of-the-art deep learning models. Nonetheless, its implications on bias in medical imaging remain unexplored. Our study investigates deep learning fairness through the lens of NC. We analyze the training dynamics of models as they approach NC when training using biased datasets, and examine the subsequent impact on test performance, specifically focusing on label bias. We find that biased training initially results in different NC configurations across subgroups, before converging to a final NC solution by memorizing all data samples. Through extensive experiments on three medical imaging datasets -- PAPILA, HAM10000, and CheXpert -- we find that in biased settings, NC can lead to a significant drop in F1 score across all subgroups. Our code is available at https://gitlab.com/radiology/neuro/neural-collapse-fairness