Enhancing Multiple Sclerosis Diagnosis with eXplainable AI

Nighat Bibi
Jane Courtney, Technological University Dublin, Ireland
Kathleen M. Curran, University College Dublin

Document Type Conference Paper

26th Irish Machine Vision and Image Processing Conference (IMVIP 2024)

https://digital-library.theiet.org/doi/10.1049/icp.2024.3308

doi:10.1049/icp.2024.3308

Abstract

Multiple sclerosis (MS) is a complex neurological disorder that requires precise diagnosis for effective treatment. This study aims to enhance MS diagnosis by integrating eXplainable Artificial Intelligence (XAI) techniques into a convolutional neural network (CNN) framework. The proposed model achieves high accuracy and provides visual explanations of its predictions. Using the Gradient-weighted Class Activation Mapping (Grad-CAM) method, it highlights the most important regions in MRI images influencing the model’s decisions, adding transparency and trust to the diagnostic process. The CNN, trained on a dataset of FLAIR MRI images, demonstrates superior performance compared to existing models, with a final accuracy of 99.36%. This work contributes to the growing field of XAI in healthcare, offering a robust and interpretable tool for MS diagnosis.