Mitigating Individual Skin Tone Bias in Skin Lesion Classification through Distribution-Aware Reweighting

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Paxton, Kuniko, Dehghani, Zeinab, Aslansefat, Koorosh, Thakker, Dhavalkumar, Papadopoulos, Yiannis
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911310361395200
author Paxton, Kuniko
Dehghani, Zeinab
Aslansefat, Koorosh
Thakker, Dhavalkumar
Papadopoulos, Yiannis
author_facet Paxton, Kuniko
Dehghani, Zeinab
Aslansefat, Koorosh
Thakker, Dhavalkumar
Papadopoulos, Yiannis
contents Skin color has historically been a focal point of discrimination, yet fairness research in machine learning for medical imaging often relies on coarse subgroup categories, overlooking individual-level variations. Such group-based approaches risk obscuring biases faced by outliers within subgroups. This study introduces a distribution-based framework for evaluating and mitigating individual fairness in skin lesion classification. We treat skin tone as a continuous attribute rather than a categorical label, and employ kernel density estimation (KDE) to model its distribution. We further compare twelve statistical distance metrics to quantify disparities between skin tone distributions and propose a distance-based reweighting (DRW) loss function to correct underrepresentation in minority tones. Experiments across CNN and Transformer models demonstrate: (i) the limitations of categorical reweighting in capturing individual-level disparities, and (ii) the superior performance of distribution-based reweighting, particularly with Fidelity Similarity (FS), Wasserstein Distance (WD), Hellinger Metric (HM), and Harmonic Mean Similarity (HS). These findings establish a robust methodology for advancing fairness at individual level in dermatological AI systems, and highlight broader implications for sensitive continuous attributes in medical image analysis.
format Preprint
id arxiv_https___arxiv_org_abs_2512_08733
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Mitigating Individual Skin Tone Bias in Skin Lesion Classification through Distribution-Aware Reweighting
Paxton, Kuniko
Dehghani, Zeinab
Aslansefat, Koorosh
Thakker, Dhavalkumar
Papadopoulos, Yiannis
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
Skin color has historically been a focal point of discrimination, yet fairness research in machine learning for medical imaging often relies on coarse subgroup categories, overlooking individual-level variations. Such group-based approaches risk obscuring biases faced by outliers within subgroups. This study introduces a distribution-based framework for evaluating and mitigating individual fairness in skin lesion classification. We treat skin tone as a continuous attribute rather than a categorical label, and employ kernel density estimation (KDE) to model its distribution. We further compare twelve statistical distance metrics to quantify disparities between skin tone distributions and propose a distance-based reweighting (DRW) loss function to correct underrepresentation in minority tones. Experiments across CNN and Transformer models demonstrate: (i) the limitations of categorical reweighting in capturing individual-level disparities, and (ii) the superior performance of distribution-based reweighting, particularly with Fidelity Similarity (FS), Wasserstein Distance (WD), Hellinger Metric (HM), and Harmonic Mean Similarity (HS). These findings establish a robust methodology for advancing fairness at individual level in dermatological AI systems, and highlight broader implications for sensitive continuous attributes in medical image analysis.
title Mitigating Individual Skin Tone Bias in Skin Lesion Classification through Distribution-Aware Reweighting
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2512.08733