Summary: An international team of scientists and clinicians developed a generative artificial intelligence framework that unmasks these previously hidden cortical lesions by analyzing existing legacy MRI scans. By synthesizing minor, sub-visual discrepancies across multiple image contrasts, the AI acts as a computational lens, pulling vital diagnostic data out of ordinary scans and revealing an entirely invisible layer of MS pathology.
Key Facts
- The Clinical Blindspot: While the newest generation of MS therapeutics developed over the past decade can slow disability progression significantly, their tracking and design have focused primarily on reducing white matter lesions because gray matter lesions were functionally invisible on standard scanners.
- The Spatial Loophole: Although individual, standalone MRI slices look normal in the gray matter, generative AI models can analyze the inter-image relationships across different scan contrasts. The AI detects tiny, sub-visual discrepancies in tissue behavior to synthesize the missing pathological landscape.
- The MMCLE Breakthrough: The research team combined multiple advanced image-processing techniques, culminating in a newly engineered protocol dubbed MMCLE (Multimodal Cortical Lesion Enhancement).
- Over 11,000 Lesions Exposed: When the MMCLE algorithm was applied to the ORATORIO trial database, it uncovered a staggering volume of hidden damage. While standard scans showed mostly white matter indicators, the AI exposed 15 to 20 previously invisible cortical lesions per patient, totaling more than 11,000 undetected lesions across the entire cohort.
- Unlocking Legacy Scan Data: Because this algorithm works perfectly on conventional, legacy MRI scans, clinics do not need to purchase multi-million dollar upgraded imaging hardware to utilize it. Doctors can immediately run old or current scans through the software to evaluate a patient’s true progression.
- A New Era for Clinical Trials: Senior author Robert Zivadinov points out that finally seeing these indicators has massive implications. It allows pharmaceutical companies to re-evaluate decades of past clinical trial data and engineer future drugs that specifically target cognitive decline in the brain’s gray matter.
Source: University at Buffalo
One of the uncomfortable truths about multiple sclerosis is that the part of the brain likely to reveal the most about the disease and how a patient will be impacted has been mostly invisible to clinicians.
It’s long been known that the gray matter of the brain plays a key role in MS disease progression and cognitive impairment, but because magnetic resonance imaging (MRI) has only been able to detect lesions in white matter, neither clinicians nor researchers have had a way to detect or monitor gray matter (cortical) lesions. And while many new drugs developed in the past decade can slow disease progression significantly, they primarily work on reducing white matter lesions.
Now, in a paper published in Communications Medicine, a University at Buffalo-led team reports that it has found a way to use artificial intelligence to reveal these otherwise invisible cortical lesions by reviewing existing MRI scans.
The significance of finally being able to see what has been known as one of the most important indicators in MS disease progression cannot be overstated, the researchers say.
“Detecting previously invisible cortical lesions on conventional legacy MRI scans has major implications for MS research and clinical care,” says Robert Zivadinov, MD, PhD, senior author on the paper, SUNY Distinguished Professor in the Department of Neurology and director of the Buffalo Neuroimaging Analysis Center (BNAC) in the Jacobs School of Medicine and Biomedical Sciences at UB. “The ability to see for the first time these previously hidden indicators of MS disease progression, including cognitive impairment and disability, is an important advance,” he says.
While the involvement of cortical lesions in MS has been known almost since the identification of MS in the late 19th century, they weren’t included on diagnostic criteria until the 21st century. And even when they were included, it was noted that their use would be greatly limited due to the current capabilities of clinical MRI.
Ongoing damage that couldn’t be seen
“We have all been very frustrated, knowing that these cortical lesions were there but not being able to see them,” says Michael G. Dwyer, PhD, first and corresponding author on the paper, associate professor of neurology and biomedical informatics in the Jacobs School and a researcher with BNAC. “There’s a lot of ongoing damage that continues to happen in MS that you won’t see with conventional MRI, but that histopathologists have been clearly demonstrating for decades on postmortem tissue.
“What this collaboration has been able to accomplish is a real success story for applying AI in the medical arena,” he continues. “We now have access to these incredibly useful data on MRI scans that were there but you couldn’t see them without using AI to pull them out. The computational methods are finally at the point where we can do this.”
The AI approaches the researchers used, building on work from the co-authors from the Netherlands, were designed to extrapolate vital information from the relationships between multiple images that can’t be seen on a single image.
The researchers combined multiple image-processing techniques, including a new one they developed called MMCLE, or multimodal cortical lesion enhancement. They then applied these techniques to MRI scans from the large, phase III FDA regulatory ORATORIO clinical trial, a study of the MS drug Ocrelizumab that included more than 700 participants.
More than 11,000 cortical lesions detected
They found that while individual images of a patient’s brain revealed mostly white matter lesions, once they applied the AI-based image processing methods to multiple different contrast images, they were able to see anywhere from 15 to 20 cortical lesions for each patient, more than 11,000 for the whole dataset.
“If you look on the original scans, you generally can’t see the cortical lesions,” says Dwyer, “but generative AI is very powerful because it can look between the scans and detect tiny differences between them. Because it sees those minor discrepancies, AI can reveal that there’s something going wrong there, that the tissue is not behaving like healthy tissue. The trained models can view multiple MRI images together and synthesize them, and synthesize what had been missing.”
Led by UB, the international research team included scientists and clinicians from academia and industry, including Genentech, which makes Ocrelizumab. Zivadinov notes the collaboration among people with such a breadth of perspectives is what contributed to their success.
“This work, which has revealed that there is so much invisible pathology in the brain, will have tremendous impact for reviewing data from past clinical trials and also for those going forward,” he says.
In addition to Zivadinov and Dwyer, UB co-authors include Niels P. Bergsland, PhD, assistant professor of neurology; Alexander Bartnik, PhD, postdoctoral researcher; and Dejan Jakimovski, MD, PhD, research adviser at BNAC.
Other co-authors are Samantha Noteboom, Menno M. Schoonheim and Martijn D. Steenwijk, all of the MS Center Amsterdam, Anatomy and Neurosciences, Vrije Universiteit Amsterdam, and Jinglan Pei and David Clayton of Genentech Inc.
Funding: The research was supported in part by Genentech.
Key Questions Answered:
A: It comes down to a limitation in contrast and resolution. While standard clinical MRI machines are highly tuned to spot bright, obvious inflammation patches within the brain’s white matter, the subtle tissue changes that occur within gray matter (the cortex) blend right into the surrounding healthy tissue on a single image. For decades, histopathologists could easily see this gray matter damage when looking at tissue samples under a microscope after a patient passed away, but living clinicians had no way to peer past the white matter barrier.
A: The AI doesn’t just look at one image in isolation; it functions like a master detective cross-referencing multiple clues. Dr. Michael G. Dwyer explains that generative AI can scan between several different image contrasts of the same brain simultaneously. By evaluating the mathematical relationships and tiny differences between these images, the AI identifies minor discrepancies showing that the gray matter tissue isn’t behaving like healthy tissue. It then uses these clues to synthesize and render the missing cortical lesions.
A: This is a massive leap forward for both current clinical care and future drug development. Because this technology works on ordinary, legacy MRI scans that hospitals already use, doctors can immediately get a much more accurate picture of a patient’s true disease progression, particularly regarding hidden indicators like disability and cognitive decline. Furthermore, it allows scientists to go back and audit data from past clinical trials to figure out why certain drugs work better than others, paving the way for a brand-new generation of MS medications designed to protect gray matter.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About this AI and multiple sclerosis research news
Author: Ellen Goldbaum
Source: University at Buffalo
Contact: Ellen Goldbaum – University at Buffalo
Image: The image is credited to Neuroscience News
Original Research: Open access.
“Quantifying cortical lesions in multiple sclerosis MRI datasets using multi-contrast post-processing and deep learning” by Michael G. Dwyer, Niels Bergsland, Alexander Bartnik, Dejan Jakimovski, Samantha Noteboom, Menno M. Schoonheim, Martijn D. Steenwijk, Jinglan Pei, David Clayton & Robert Zivadinov. Communications Medicine
DOI:10.1038/s43856-026-01683-7
Abstract
Quantifying cortical lesions in multiple sclerosis MRI datasets using multi-contrast post-processing and deep learning
Background
Multiple sclerosis (MS) is a chronic neurological disease affecting both white and gray matter of the central nervous system. Despite the well-established involvement of cortical lesions in MS, feasibility limitations in their visualization on typical magnetic resonance imaging (MRI) protocols prevent their evaluation in nearly all clinical trials. Recently, several post-processing methods, including synthetic contrasts and artificial intelligence (AI)-based approaches, have shown potential for enhancing cortical lesion detection on conventional MRI data. These methods have the potential to reanalyze existing clinical-trial data to answer key mechanistic questions about both MS development and about treatment effects.
Methods
We sought to evaluate the feasibility of combining and extending existing methods into a unified framework for analysis using the data from the large, multicenter, phase 3 ORATORIO trial (full n = 732, age=44.6 ± 8.0; development subset n = 80, age=46.6 ± 7.1). We specifically evaluated three of the most promising of them – fluid-attenuated inversion recovery squared (FLAIR2), T1/T2 ratio, and artificial intelligence-derived double inversion recovery (AI-DIR) – and introduced a new combined contrast called multi-modal cortical lesion enhanced (MMCLE). We also harnessed transformer-based semantic segmentation to improve automated detection and delineation of these lesions.
Results
At baseline, we detected 14.8 + /−20.72 lesions per participant, with 86.0% true positive rate and 8.4% false positive rate across subjects for blinded MMCLE, using simultaneous review of all contrasts as the reference. High reproducibility was observed across field strengths and acquisition types (ICC 88.8-92.5%).
Conclusions
We confirmed that cortical lesions can be clearly visualized and quantified with these methods. Using deep learning, we also confirmed that the simultaneous use of multiple contrasts improves quantification.
