Bias Analysis and Data Augmentation Strategies in Skin Lesion Classification

Thiago Meneses Lopes; Martin Heckmann; Marcelo Ponciano da Silva; Pedro Henrique Bugatti; Priscila Tiemi Maeda Saito

doi:10.22456/2175-2745.150974

Authors

Thiago Meneses Lopes Universidade Federal de São Carlos (UFSCar)
Martin Heckmann Universidade Federal de São Carlos (UFSCar)
Marcelo Ponciano da Silva Instituto Federal do Triângulo Mineiro (IFTM)
Pedro Henrique Bugatti Universidade Federal de São Carlos (UFSCar) https://orcid.org/0009-0005-5558-5231
Priscila Tiemi Maeda Saito Universidade Federal de São Carlos (UFSCar)

DOI:

https://doi.org/10.22456/2175-2745.150974

Keywords:

data augmentation, classification, skin lesion, computer vision, data bias

Abstract

Early detection of skin cancer is vital for effective treatment and improving patient recovery. In recent years, a growing number of computer vision studies have been developed to aid in diagnosis, drawing significant attention from researchers. However, challenges still persist, such as data imbalance and the lack of comprehensive datasets. Additionally, limited research has focused on how variations in skin tone across different populations affect the performance of models in skin lesion classification. This study seeks to create a more effective approach to address data biases in lesion classification across diverse skin tones. We initially explored several data augmentation techniques, employing traditional feature extractors for image analysis. For classification, models such as k-Nearest Neighbor, Random Forest, and Support Vector Machine were used. This study focused on two well-known and publicly available skin lesion datasets: HAM10000 and PAD-UFES-20, both of which have significant class imbalances. Further experiments were conducted to assess potential biases, with the Individual Typology Angle (ITA) metric applied to evaluate the skin tone distribution within the datasets.

Downloads

Download data is not yet available.

References

[1] CHEN, J. Y. et al. Skin cancer diagnosis by lesion, physician, and examination type: a systematic review and meta-analysis. JAMA Dermatology, 2025.

[2] SALINAS, M. P. et al. A systematic review and meta-analysis of artificial intelligence versus clinicians for skin cancer diagnosis. NPJ Digital Medicine, v. 7, n. 1, p. 125, 2024.

[3] ALDRIDGE, R. B.; MAXWELL, S. S.; REES, J. L. Dermatology undergraduate skin cancer training: a disconnect between recommendations, clinical exposure and competence. BMC Medical Education, v. 12, n. 1, p. 27, 2012.

[4] KINYANJUI, N. M. et al. Estimating skin tone and effects on classification performance in dermatology datasets. 2019. ArXiv preprint arXiv:1910.13268.

[5] PEREZ, F. et al. Data augmentation for skin lesion analysis. In: International Workshop on Computer-Assisted and Robotic Endoscopy (CARE). Cham: Springer International Publishing, 2018. p. 303–311.

[6] RASHID, H.; TANVEER, M. A.; KHAN, H. A. Skin lesion classification using GAN-based data augmentation. In: 2019 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). Berlin, Germany: IEEE, 2019. p. 916–919.

[7] PHAM, T.-C. et al. Deep CNN and data augmentation for skin lesion classification. In: Asian Conference on Intelligent Information and Database Systems (ACIIDS). Cham: Springer International Publishing, 2018. p. 573–582.

[8] BOZKURT, F. Skin lesion classification on dermatoscopic images using effective data augmentation and pre-trained deep learning approach. Multimed. Tools Appl., v. 82, n. 12, p. 18985–19003, 2023.

[9] GALDRAN, A. et al. Data-driven color augmentation techniques for deep skin image analysis. 2017. ArXiv preprint arXiv:1703.03702.

[10] KINYANJUI, N. M. et al. Fairness of classifiers across skin tones in dermatology. In: International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI). Cham: Springer International Publishing, 2020. p. 320–329.

[11] DANESHJOU, R. et al. Disparities in dermatology AI performance on a diverse, curated clinical image set. Science Advances, v. 8, n. 31, p. eabq6147, 2022.

[12] BARROS, L.; CHAVES, L.; AVILA, S. Assessing the generalizability of deep neural networks-based models for black skin lesions. In: Iberoamerican Congress on Pattern Recognition (CIARP). Cham: Springer Nature Switzerland, 2023. p. 1–14.

[13] CORREA-MEDERO, R. L.; PATEL, B.; BANERJEE, I. Adversarial debiasing techniques towards ‘fair’ skin lesion classification. In: 2023 11th International IEEE/EMBS Conference on Neural Engineering (NER). Florence, Italy: IEEE, 2023. p. 1–4.

[14] TSCHANDL, P.; ROSENDAHL, C.; KITTLER, H. The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions. Scientific Data, v. 5, n. 1, p. 1–9, 2018.

[15] PACHECO, A. G. C. et al. PAD-UFES-20: A skin lesion dataset composed of patient data and clinical images collected from smartphones. Data in Brief, v. 32, p. 106221, 2020.

[16] CORBIN, A.; MARQUES, O. Exploring strategies to generate Fitzpatrick skin type metadata for dermoscopic images using individual typology angle techniques. Multimedia Tools and Applications, v. 82, n. 15, p. 23771–23795, 2023.

[17] CHARDON, A.; CRETOIS, I.; HOURSEAU, C. Skin colour typology and suntanning pathways. International Journal of Cosmetic Science, v. 13, n. 4, p. 191–208, 1991.