AI makes the invisible visible in multiple sclerosis. Researchers used AI-enhanced image processing on standard MRI scans from over 700 MS patients and detected more than 10,000 cortical lesions that conventional scans had missed entirely. The breakthrough could reshape how disease progression is tracked and how future MS clinical trials are designed.
AI makes the invisible visible in multiple sclerosis
Cortical lesions — found in the brain's gray matter — are strongly linked to disability, cognitive decline, and disease progression in MS, but standard MRI scans have long struggled to detect them. Now, researchers at the University at Buffalo have used AI-enhanced post-processing techniques on conventional MRI data from the phase 3 ORATORIO trial of ocrelizumab, uncovering lesions that were essentially hidden in plain sight.
Across 732 patients, the AI approach detected over 10,000 cortical lesions — roughly 14 per patient — that standard scans missed. The team's newly developed technique, multimodal cortical lesion enhancement (MMCLE), led the pack, correctly identifying 86% of cortical lesions with a low false-positive rate of just 8.4%, and proved reproducible across different MRI scanners and protocols.
By the Numbers
Why it matters: This technology could allow researchers to reanalyze existing clinical trial datasets and design better future MS trials. While routine clinical use is still down the road, AI-based cortical lesion detection may eventually help identify high-risk patients and catch disease activity that conventional MRI misses.
