Researchers from the University at Buffalo, leading an international collaboration, have successfully applied artificial intelligence to detect previously invisible cortical lesions in patients with multiple sclerosis using standard MRI scans. The findings, published in Communications Medicine, address a longstanding limitation in neuroimaging where conventional MRI technology has historically been unable to visualize gray matter damage, which is closely tied to disease progression and cognitive decline. The AI framework, developed by the Buffalo Neuroimaging Analysis Center in partnership with researchers in the Netherlands and Genentech, leverages generative machine learning to analyze relationships across multiple MRI contrast images. Unlike traditional methods that rely on single scans, the system employs a novel technique called multimodal cortical lesion enhancement to identify microscopic tissue discrepancies invisible to conventional viewing. When applied to imaging data from the Phase III ORATORIO clinical trial, which evaluated the MS drug Ocrelizumab across more than 700 participants, the model identified an average of 15 to 20 cortical lesions per patient, totaling over 11,000 lesions across the dataset. Prior to this advancement, cortical lesions could only be confirmed through postmortem histopathology and were largely excluded from clinical monitoring and diagnostic criteria due to technical imaging constraints. The ability to extract these markers from existing legacy scans transforms available historical data. Senior author Robert Zivadinov noted that the discovery of extensive hidden pathology will substantially improve the evaluation of both past and ongoing clinical trials, while enabling more precise tracking of neurodegenerative changes. First author Michael G. Dwyer emphasized that computational methods have now matured sufficiently to bridge the gap between visible white matter damage and the subtle gray matter degeneration that drives long-term disability. This breakthrough establishes a scalable, non-invasive imaging protocol that requires no new hardware. By repurposing standard clinical MRI sequences, the AI-driven approach lowers barriers to early disease detection and long-term monitoring. The methodology is expected to accelerate neuroimmunology research, enhance treatment efficacy assessments for emerging MS therapeutics, and provide clinicians with a previously unattainable window into cortical neurodegeneration.