Researchers at the University at Buffalo have developed an artificial intelligence method to detect cortical lesions in multiple sclerosis (MS) patients using existing MRI scans. By analyzing relationships between multiple contrast images, this AI-driven approach reveals invisible gray matter damage, providing a new metric for tracking disease progression and cognitive impairment, according to a study published in Communications Medicine.
How Does AI Reveal Invisible MS Lesions?
Conventional MRI technology has long been limited to identifying white matter lesions, leaving the gray matter—a critical indicator of MS progression—largely hidden from clinicians. According to Michael G. Dwyer, PhD, the paper’s first and corresponding author, the new computational methods synthesize data from multiple MRI images to identify tissue abnormalities that appear healthy on a single scan.
The research team utilized a technique called multimodal cortical lesion enhancement (MMCLE). By processing these images through deep learning models, the researchers identified between 15 and 20 cortical lesions per patient that were previously undetectable. Across the study’s dataset, which included more than 700 participants from the phase III ORATORIO clinical trial of Ocrelizumab, the AI detected over 11,000 cortical lesions.
While cortical lesions were identified as a feature of MS in the late 19th century, they were not formally included in diagnostic criteria until the 21st century due to the technological limitations of standard clinical MRI.
Why Is Seeing Gray Matter Damage Important for MS Care?
The ability to visualize cortical lesions transforms how clinicians and researchers monitor disease progression. Robert Zivadinov, MD, PhD, senior author of the paper and director of the Buffalo Neuroimaging Analysis Center (BNAC), states that this advancement has major implications for both research and clinical care. Because these lesions are linked to cognitive impairment and disability, monitoring them provides a more accurate picture of how a patient’s disease is evolving.
Historically, drug development has focused on white matter lesions. By applying this AI-based processing to legacy MRI data, researchers can now re-evaluate past clinical trials. This allows for a more comprehensive understanding of how current treatments impact the brain’s gray matter, potentially influencing future therapeutic strategies.
What Are the Future Trends in MS Diagnostics?
The integration of AI into medical imaging signals a shift toward utilizing existing clinical data to extract deeper insights. As noted by Dwyer, the computational methods have finally reached a level of maturity where they can identify the minor discrepancies between MRI scans that indicate damaged tissue. This suggests that future diagnostic protocols may rely increasingly on AI-augmented image processing to provide earlier and more precise assessments of neurodegenerative diseases.
Clinicians and researchers looking to apply these methods can look to the MMCLE technique described in the Communications Medicine report, which demonstrates how multi-contrast processing can extract vital data from standard, non-specialized MRI datasets.
Frequently Asked Questions
Why haven’t cortical lesions been visible on MRI scans before?
Conventional MRI technology is optimized to detect white matter lesions. According to the study authors, cortical lesions were known to exist through postmortem histopathology but remained invisible on standard clinical scans until AI-based processing could synthesize data across multiple contrast images.
What does this mean for MS patients?
This development allows for better tracking of disease progression. By identifying previously hidden cortical lesions, doctors may gain a clearer understanding of the factors contributing to cognitive impairment and disability in individual patients.
Was this study limited to one specific drug?
The team applied their AI techniques to data from the phase III ORATORIO clinical trial, which studied the drug Ocrelizumab. However, the researchers emphasize that this method can be applied to other datasets, including past clinical trials and future research, to reveal invisible pathology.
To stay updated on the latest breakthroughs in neurology and medical imaging, subscribe to our newsletter or explore our archive of clinical research reports. Have questions about how AI is changing diagnostic medicine? Join the conversation in the comments section below.
