Detecting Melanoma Fairly: Skin Tone Detection and Debiasing for Skin Lesion Classification

Bevan, Peter J.; Atapour-Abarghouei, Amir

Abstract

Convolutional Neural Networks have demonstrated human-level performance in the classification of melanoma and other skin lesions, but evident performance disparities between differing skin tones should be addressed before widespread deployment. In this work, we propose an efficient yet effective algorithm for automatically labelling the skin tone of lesion images, and use this to annotate the benchmark ISIC dataset. We subsequently use these automated labels as the target for two leading bias ‘unlearning’ techniques towards mitigating skin tone bias. Our experimental results provide evidence that our skin tone detection algorithm outperforms existing solutions and that ‘unlearning’ skin tone may improve generalisation and can reduce the performance disparity between melanoma detection in lighter and darker skin tones.

Citation

Bevan, P. J., & Atapour-Abarghouei, A. (2022, December). Detecting Melanoma Fairly: Skin Tone Detection and Debiasing for Skin Lesion Classification. Presented at DART: MICCAI Workshop on Domain Adaptation and Representation Transfer

• Presentation Conference Type: Conference Paper (published)
• Conference Name: DART: MICCAI Workshop on Domain Adaptation and Representation Transfer
• Acceptance Date: Aug 15, 2022
• Online Publication Date: Sep 15, 2022
• Publication Date: 2022
• Deposit Date: Sep 20, 2022
• Publicly Available Date: Sep 16, 2023
• Print ISSN: 0302-9743
• Volume: 13542
• Pages: 1-11
• Series Title: Lecture Notes in Computer Science
• Series ISSN: 0302-9743,1611-3349
• Book Title: DART 2022: Domain Adaptation and Representation Transfer
• ISBN: 9783031168512
• DOI: https://doi.org/10.1007/978-3-031-16852-9_1
• Public URL: https://durham-repository.worktribe.com/output/1135904