Open Access

ARTICLE

Advancing Radiological Dermatology with an Optimized Ensemble Deep Learning Model for Skin Lesion Classification

Adeel Akram¹, Tallha Akram², Ghada Atteia^3,*, Ayman Qahmash⁴, Sultan Alanazi⁵, Faisal Mohammad Alotaibi⁵

1 Department of Electrical and Computer Engineering, COMSATS University Islamabad, Wah Campus, Wah Cantonment, 47040, Pakistan
2 Department of Information Systems, College of Computer Engineering and Sciences, Prince Sattam bin Abdulaziz University, Al-Kharj, 16273, Saudi Arabia
3 Department of Information Technology, College of Computer and Information Sciences, Princess Nourah bint Abdulrahman University, P.O. Box 84428, Riyadh, 11671, Saudi Arabia
4 Informatics and Computer Systems Department, King Khalid University, Abha, 62217, Saudi Arabia
5 Department of Computer Science, College of Computer Engineering and Sciences, Prince Sattam bin Abdulaziz University, Al-Kharj, 16273, Saudi Arabia

* Corresponding Author: Ghada Atteia. Email:

(This article belongs to the Special Issue: Machine Learning and Deep Learning-Based Pattern Recognition)

Computer Modeling in Engineering & Sciences 2025, 145(2), 2311-2337. https://doi.org/10.32604/cmes.2025.069697

Received 28 June 2025; Accepted 09 October 2025; Issue published 26 November 2025

Abstract

Advancements in radiation-based imaging and computational intelligence have significantly improved medical diagnostics, particularly in dermatology. This study presents an ensemble-based skin lesion classification framework that integrates deep neural networks (DNNs) with transfer learning, a customized DNN, and an optimized self-learning binary differential evolution (SLBDE) algorithm for feature selection and fusion. Leveraging computational techniques alongside medical imaging modalities, the proposed framework extracts and fuses discriminative features from multiple pre-trained models to improve classification robustness. The methodology is evaluated on benchmark datasets, including ISIC 2017 and the Argentina Skin Lesion dataset, demonstrating superior accuracy, precision, and F1-score in melanoma detection. The proposed method achieved a classification accuracy of 98.5%, evaluated using an LSVM classifier on the Argentina Skin Lesion dataset, underscoring the robustness of the proposed methodology. The proposed approach offers a scalable and computationally efficient solution for automated skin lesion classification, thereby contributing to improved clinical decision-making and enhanced patient outcomes. By aligning artificial intelligence with radiation-based medical imaging and bioinformatics, this research advances dermatological computer-aided diagnosis (CAD) systems, minimizing misclassification rates and supporting early skin cancer detection. The proposed approach provides a scalable and computationally efficient solution for automated skin lesion analysis, contributing to improved clinical decision-making and enhanced patient outcomes.

---

Keywords

---