Diagnosis of Multiple Sclerosis from Magnetic Resonance Imaging Using Convolutional Neural Networks

Abstract

Multiple Sclerosis (MS) is a chronic neurodegenerative disease caused by the demyelination of nerve fibers in the central nervous system. Early and accurate diagnosis is critical for effective disease management. Magnetic Resonance Imaging (MRI) plays a key role in detecting white matter lesions associated with MS. In recent years, Convolutional Neural Networks (CNNs) have emerged as powerful tools for analyzing medical images, offering high accuracy in detecting MS-related abnormalities. This review provides a comprehensive and comparative analysis of studies conducted from 2023, focusing on the application of CNN-based approaches for the diagnosis of MS using MRI. This paper highlights recent methodological innovations, evaluates their diagnostic performance, and identifies existing research gaps. The selection of this time window aims to reflect the latest advancements in deep learning techniques and their implementation in clinical neuroimaging. Articles are selected based on specific inclusion criteria from Web of Science, PubMed, and IEEE Xplore databases. The results indicate that CNNs continue to demonstrate high efficiency in feature extraction and lesion identification, making them suitable for automated MS diagnosis. The review also suggests directions for future studies to address current limitations and enhance clinical applicability.