Citation

Nabi, Md Serajun and Islam, Md Rashidul and Gibba, Bakary and Mohammadi, Raihana and Touhami, Meriem and Ahmad Fauzi, Mohammad Faizal (2025) Skin Lesion Classification with Explainable AI to Enhancing Dermatological Diagnosis. In: 2025 Multimedia University Engineering Conference, MECON 2025, 21 July 2025 - 23 July 2025, Cyberjaya, Malaysia.

Text 475.pdf - Published Version Restricted to Repository staff only Download (4MB)

Abstract

Melanoma is one of the most dangerous forms of cancer in the world and is particularly common among skin cancers. However, conventional diagnostic methods remain subjective, slow, and prone to error and early detection significantly improves patient survival. To solve this problem, this study developed three deep learning models (CNN, DenseNet201, and UNet) to classify skin lesions from dermoscopic images. Advanced techniques such as attention modules (SE, CBAM), data augmentation (Mixup), and explainable AI (Grad-CAM) were implemented to increase model accuracy and interpretability. Among these, the UNet model demonstrated the highest accuracy of 95%, followed by DenseNet201 (93%) and CNN (90%). Additionally, Grad-CAM demostrates clear visual explanations for the models’ decisions, improving clinical trust. The results highlight the effectiveness of combining deep learning models with explainable methods. This approach provides a reliable, accurate, and transparent tool for automatic melanoma diagnosis, potentially enhancing clinical decision-making in dermatology.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Skin cancer detection, Melanoma classification, explainable AI, Grad-CAM, atypical nevus, common nevus, dermoscopic image
Subjects:	R Medicine > RC Internal medicine > RC71-78.7 Examination. Diagnosis
Divisions:	Faculty of Artificial Intelligence & Engineering (FAIE)
Depositing User:	Ms Suzilawati Abu Samah
Date Deposited:	19 Mar 2026 01:16
Last Modified:	19 Mar 2026 01:16
URII:	http://shdl.mmu.edu.my/id/eprint/15493

Downloads

Edit (login required)