← Back to Publications List

AI-Powered Skin Disease Detection Using Multimodal Imaging and Explainable AI

Students & Supervisors

Student Authors
Nishat Tasnim
Master of Science in Computer Science, FST
Supervisors
Firoz Ahmed
Professor, Faculty, FST

Abstract

Millions of people worldwide suffer from skin dis eases, and they require timely and accurate diagnosis. COVID 19 has also complicated dermatological conditions due to virus induced complications or drug reactions. Traditional diagnosis is dependent on dermatologists and subjective in nature. This work proposes AI-based skin disease detection that combines Explainable AI (XAI) with multimodal imaging—dermoscopic, clinical, thermal, and RGB—to improve the accuracy of clas sification and explainability. The HAM10000 dataset was used for dermoscopic images, while clinical and thermal images were simulated from the same dataset to emulate multimodal input. Using state-of-the-art deep learning models and attention inspired feature fusion, the system effectively combines data modalities. Interpretability is ensured with Local Interpretable Model-Agnostic Explanations (LIME). Evaluation on benchmark and simulated clinical datasets demonstrates improved accuracy and diagnostic confidence, promoting practical deployment in dermatology.

Keywords

AI Skin Disease Detection Multimodal Imaging Explainable AI Deep Learning

Publication Details

  • Type of Publication:
  • Conference Name: 2025 International Conference on Quantum Photonics, Artificial Intelligence, and Networking (QPAIN)
  • Date of Conference: 31/07/2025 - 31/07/2025
  • Venue: Rangpur
  • Organizer: IEEE Photonics Society Bangladesh Chapter
View on Publisher