Using Machine Learning Methods to Improve the Diagnosis and Classification of Skin Diseases from Dermatoscopic Images

Viktoriya Lylikova, Mikhail Bezriadin

Abstract


Skin diseases are the most common diseases worldwide. At best, they cause physical discomfort, and in severe cases, they can lead to skin cancer. Dermatoscopy images offer a non-invasive and cost-effective method for investigating skin abnormalities. Deep learning models have shown promising results in computer-aided diagnosis by automatically extracting features from medical images. However, the performance of these models largely depends on the quality and balance of the training dataset. This study demonstrates the feasibility of classifying dermatoscopy images of skin lesions into eight classes for diagnosis. Pre-trained neural networks DenseNet201, MobileNetV2, and ResNet50 were used to solve this problem. The dataset underwent several preprocessing steps, including data cleaning, image resizing, and class balancing. The main architectural parameters of the proposed neural networks are described. A comparative analysis of the performance of the developed algorithms is conducted. The results of the study show that the use of deep neural networks allows achieving an accuracy of 95.5%.

Full Text:

PDF (Russian)

References


Sivodedova N.A., Karyakin N.N., Shlivko I.L. Modern methods for detecting malignant skin neoplasms, including the use of mobile applications and artificial intelligence: a literature review // Medical informatics: theory and practice, ORGZDRAV: news, opinions, training. Bulletin of the Higher School of Medicine and Health. Vol. 10, No. 2, 2024, pp. 78-93.

Chebotarev V.V., Khismatullina Z.R., Zakirova Yu.A. Some aspects of the epidemiology and diagnostics of malignant skin neoplasms // Creative surgery and oncology. 2020. Vol. 10, No. 1. P. 65–73

Malignant neoplasms in Russia in 2021 (morbidity and mortality): collection / edited by A.D. Kaprin, V.V. Starinsky, A.O. Shakhzadova. Moscow: P.A. Herzen Medical Research Institute of Oncology – branch of the National Medical Research Center of Radiology of the Ministry of Health of the Russian Federation, 2022. 252 p.

Mahmoud, Nourelhoda & Soliman, Ahmed. (2024). Early automated detection system for skin cancer diagnosis using artificial intelligent techniques. Scientific Reports. 14. 10.1038/s41598-024-59783-0.

Sara, Umme & Das, Utpol & Sikder, Juel & Chowdhury, Sudipta. (2025). Automated Skin Cancer Classification and Detection Using Convolutional Neural Networks and Dermoscopy Images. Procedia Computer Science. 252. 108-117. 10.1016/j.procs.2024.12.012.

Philip, Felix & Ajesh, F. & Izdrui, Diana & Craciun, Maria-Daniela & Geman, Oana. (2022). Automated Detection of Nonmelanoma Skin Cancer Based on Deep Convolutional Neural Network. Journal of Healthcare Engineering. 2022. 1-15. 10.1155/2022/6952304.

Zhang, Long & Gao, Hong & Zhang, Jianhua & Badami, Benjamin. (2020). Optimization of the Convolutional Neural Networks for Automatic Detection of Skin Cancer. Open Medicine. 15. 27-37. 10.1515/med-2020-0006.

Öztürk, Şaban & Özkaya, Umut. (2020). Skin Lesion Segmentation with Improved Convolutional Neural Network. Journal of Digital Imaging. 33. 958-970. 10.1007/s10278-020-00343-z.

Yang, Cheng-Hong & Ren, Jai-Hong & Huang, Hsiu-Chen & Chuang, Li-Yeh & Chang, Po-Yin. (2021). Deep Hybrid Convolutional Neural Network for Segmentation of Melanoma Skin Lesion. Computational Intelligence and Neuroscience. 2021. 1-15. 10.1155/2021/9409508.

Sikandar, Shahbaz & Mahum, Rabbia & Ragab, Adham & Yildirim Yayilgan, Sule & Shaikh, Sarang. (2023). SCDet: A Robust Approach for the Detection of Skin Lesions. Diagnostics. 13. 10.3390/diagnostics13111824.

Afza, Farhat & Sharif, Muhammad & Khan, M. & Tariq, Usman & Yong, Hwan-Seung & Cha, Jaehyuk. (2022). Multiclass Skin Lesion Classification Using Hybrid Deep Features Selection and Extreme Learning Machine. Sensors. 22. 10.3390/s22030799.

Haenssle, Holger & Fink, Christine & Schneiderbauer, R & Toberer, Ferdinand & Buhl, Timo & Blum, A & Kalloo, A & Hassen, A & Thomas, Luc & Enk, A & Uhlmann, Lorenz. (2018). Man against Machine: Diagnostic performance of a deep learning convolutional neural network for dermoscopic melanoma recognition in comparison to 58 dermatologists. Annals of oncology : official journal of the European Society for Medical Oncology. 29. 10.1093/annonc/mdy166.

Marchetti, Michael & Codella, Noel & Dusza, Stephen & Gutman, David & Helba, Brian & Kalloo, Aadi & Mishra, Nabin Kumar & Carrera, Cristina & Celebi, M. Emre & DeFazio, Jennifer & Jaimes, Natalia & Marghoob, Ashfaq & Quigley, Elizabeth & Scope, Alon & Yélamos, Oriol & Halpern, Allan. (2017). Results of the 2016 International Skin Imaging Collaboration International Symposium on Biomedical Imaging challenge: Comparison of the accuracy of computer algorithms to dermatologists for the diagnosis of melanoma from dermoscopic images. Journal of the American Academy of Dermatology. 78. 10.1016/j.jaad.2017.08.016.

Kozachok A.V., Spirin A.A., Samovarov O.I., Kozachok E.S. Application of machine learning models for multi-class classification of dermatoscopic images of skin neoplasms. Proceedings of the Institute for System Programming of the Russian Academy of Sciences. 2024;36(5):241-252.

Mikhailusov A. V. Using convolutional neural networks for skin cancer diagnostics // VSTU Bulletin. 2023, pp. 20-26.

Iqball, Talib & Wani, M. (2023). Weighted ensemble model for image classification. International Journal of Information Technology. 15. 10.1007/s41870-022-01149-8.

Müller, Dominik & Soto-Rey, Iñaki & Kramer, Frank. (2022). An Analysis on Ensemble Learning optimized Medical Image Classification with Deep Convolutional Neural Networks. 10.48550/arXiv.2201.11440.

Skin Cancer ISIC. The skin cancer data Режим доступа - https://www.kaggle.com/datasets/nodoubttome/skin-cancer9-classesisic


Refbacks

  • There are currently no refbacks.


Abava  Кибербезопасность Monetec 2026 СНЭ

ISSN: 2307-8162