N. qMRI-Based Classification of Multiple Sclerosis Lesions: A Machine Learning Tool for Reducing the use of Contrast Injections

Routine monitoring of Multiple Sclerosis (MS) heavily relies on Gadolinium-based contrast agents (GBCAs) to identify active brain lesions and assess disease progression. However, the frequent and repeated injection of these contrast agents raises significant safety concerns, including long-term gadolinium retention in the brain, potential renal toxicity, and allergic reactions, while also increasing healthcare costs and patient discomfort. Motivated by the critical need for safer diagnostic alternatives, this project focuses on developing a completely non-invasive, contrast-free methodology to classify MS lesion activity.

By leveraging multi-parametric quantitative MRI (qMRI) maps—capturing absolute tissue properties such as relaxation times, myelin water fraction, and diffusion—combined with advanced radiomic feature extraction and machine learning algorithms, the research aims to identify active demyelination and inflammation based purely on intrinsic microstructural tissue changes. This holistic approach seeks to replace traditional contrast-enhanced scans, offering a safer, objective, and highly personalized tool for continuous MS management.