Analysis of ResNet50 Model Response to Skin Tone Variations in Medical Image-Based Skin Disease Classification
DOI:
https://doi.org/10.30871/jaic.v10i3.12794Keywords:
Deep learning, Fitzpatrick skin type, ResNet50, skin disease classification, skin tone biasAbstract
Skin disease classification with deep learning has shown promising performance, however many models are primarily trained on datasets featuring light skin tones, which raises question about their effectiveness across a variety of akin types. This study analyses the response of a ResNet50 model based on transfer learning when faced with different skin tones in order to classifying skin disease using medical images. The model was trained on the HAM100000 which categorized into three classes: benign, malignant, and non-neoplastic. A bias analysis was then performed using the Fitzpatrick 17k dataset. The model demonstrated an overall accuracy of 70.85%, a precision rate of 74.03%, and a recall rate of 65.51%. Further analysis showed that the model had a consistent pattern of predicting malignant cases, which increased with darker skin tones, rising from 54% to 68.3%. To mitigate this issue, a threshold tuning approach was applied. After mitigation, the model achieved an accuracy of 74%, a weighted F1-score of 76%, dan a macro F1-score of 55%. Fairness evaluation after mitigation showed tha the proportion of malignant predictions increased from 56,3% in FST I to 69,9% in FST VI. These findings suggest that threshold tuning can improve classification performance and partially reduce bias intensity.
Downloads
References
[1] A. Moloo, “Recognizing neglected skin diseases: WHO publishes pictorial training guide,” who.int.
[2] D. Li et al., “Worldwide trends and future projections of fungal skin disease burden: a comprehensive analysis from the Global Burden of Diseases study 2021,” Front. Public Health, vol. 13, Jun. 2025, doi: 10.3389/fpubh.2025.1580221.
[3] N. Annalakshmi and S. Umarani, “SkinProNet: An attention-based deep learning system for skin disease classification and segmentation,” Elsevier, vol. 26, Oct. 2025, doi: 10.1016/j.simpa.2025.100798.
[4] S. Sharma, R. Mittal, N. Goyal, S. B. Goyal, and C. Verma, “Skin disease diagnostics through federated transfer learning on heterogeneous data,” Sci. Rep., vol. 16, Jan. 2026, doi: 10.1038/s41598-025-31730-7.
[5] P. N. Srinivasu, J. G. Sivasai, M. F. Ijaz, A. K. Bhoi, W. Kim, and J. J. Kang, “Classification of Skin Disease Using Deep Learning Neural Networks with MobileNet V2 and LSTM,” Sensors, vol. 21, no. 8, Apr. 2021, doi: 10.3390/s21082852.
[6] P. Tschandl, C. Rosendahl, and H. Kittler, “Data descriptor: The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions,” Sci. Data, vol. 5, Aug. 2018, doi: 10.1038/sdata.2018.161.
[7] M. Groh et al., “Deep learning-aided decision support for diagnosis of skin disease across skin tones,” Nat. Med., vol. 30, pp. 573–583, Feb. 2024, doi: 10.1038/s41591-023-02728-3.
[8] K. Nijjer et al., “Adapting Large Language Models to Mitigate Skin Tone Biases in Clinical Dermatology Tasks: A Mixed-Methods Study,” Electrical Engineering and Systems Science, Oct. 2025, [Online]. Available: http://arxiv.org/abs/2510.00055
[9] K. Mader, “Skin Cancer MNIST: HAM10000,” Kaggle.
[10] M. Farabi, “fitzpatrick 17k tonewise splitted,” Kaggle.
[11] A. Asriani, N. Lapatta, D. Nugraha, A. Amriana, and W. Wirdayanti, “Implementation of ResNet-50-Based Convolutional Neural Network For Mobile Skin Cancer Classification,” Journal of Applied Informatics and Computing (JAIC), vol. 9, no. 4, pp. 1569–1577, Jul. 2025, [Online]. Available: http://jurnal.polibatam.ac.id/index.php/JAIC
[12] F. Ritan and A. Chandra, “Analisis Perbandingan Kinerja Model CNN Resnet-50, VGG19 dan Mobilenet dalam Klasifikasi Penyakit pada Tanaman Mete,” JURNAL LOCUS: Penelitian & Pengabdian, vol. 4, no. 8, pp. 7903–7918, Aug. 2025, [Online]. Available: https://locus.rivierapublishing.id/index.php/jl
[13] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, IEEE Computer Society, 2016, pp. 770–778. doi: 10.1109/CVPR.2016.90.
[14] W. Xu, Y. L. Fu, and D. Zhu, “ResNet and its application to medical image processing: Research progress and challenges,” Comput. Methods Programs Biomed., vol. 240, Jun. 2023, doi: 10.1016/j.cmpb.2023.107660.
[15] A. Sihabillah, A. Tholib, and I. I. Basit, “OPTIMASI Model ResNet50 untuk Klasifikasi Sampah,” INDEXIA: Informatic and Computational Intelligent Journal, vol. 06, no. 02, pp. 102–111, Apr. 2025.
[16] A. Setiawan, A. ndruru, R. Rosnelly, and A. R. Zai, “Analisis Pengaruh Fine-Tuning pada Model ResNet-50 untuk Deteksi Multikategori Penyakit Mata Berdasarkan Citra Fundus Retina 1,” Remik: Riset dan E-Jurnal Manajemen Informatika Komputer, vol. 10, no. 1, Mar. 2026, doi: 10.33395/remik.v10i1.15047.
[17] R. Zhang, Y. Zhu, Z. Ge, H. Mu, D. Qi, and H. Ni, “Transfer Learning for Leaf Small Dataset Using Improved ResNet50 Network with Mixed Activation Functions,” Forests, vol. 13, no. 12, Dec. 2022, doi: 10.3390/f13122072.
[18] N. Kinyanjui et al., “Estimating Skin Tone and Effects on Classification Performance in Dermatology Datasets,” Oct. 2019, [Online]. Available: http://arxiv.org/abs/1910.13268
[19] M. Hort, Z. Chen, J. M. Zhang, M. Harman, and F. Sarro, “Bias Mitigation for Machine Learning Classifiers: A Comprehensive Survey,” ACM Journal on Responsible Computing, vol. 1, no. 2, pp. 1–52, Jun. 2024, doi: 10.1145/3631326.
[20] T. Y. Lin, P. Goyal, R. Girshick, K. He, and P. Dollar, “Focal Loss for Dense Object Detection,” in Proceedings of the IEEE International Conference on Computer Vision, Institute of Electrical and Electronics Engineers Inc., Dec. 2017, pp. 2999–3007. doi: 10.1109/ICCV.2017.324.
[21] D. Minatel, A. Parmezan, N. Santos, M. Curi, and A. Lopes, “A DIF-Driven Threshold Tuning Method for Improving Group Fairness,” in Proceedings of the ACM Symposium on Applied Computing, Association for Computing Machinery, Apr. 2025. doi: 10.1145/3672608.3707875.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Made Ireina Dwiandra Divayanti, I Gusti Ngurah Lanang Wijayakusuma
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License (Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) ) that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
