A team of researchers at University College London (UCL) has developed a new artificial intelligence (AI) tool that could revolutionize how doctors monitor and treat multiple sclerosis (MS). The tool, named MindGlide, uses advanced AI techniques to analyze brain MRI scans, identifying signs of damage and changes like brain shrinkage or lesions—often faster and more accurately than traditional methods.
MS is a chronic condition where the immune system attacks the brain and spinal cord, leading to problems with movement, sensation, and cognition. In the UK alone, it affects over 130,000 people and costs the NHS more than £2.9 billion each year.
A Smarter Way to Read Brain Scans
While MRI scans are essential for diagnosing and tracking MS, interpreting them typically requires highly trained specialists and multiple types of scans. This process can be time-consuming and isn’t always feasible during routine hospital visits. That’s where MindGlide comes in.
Instead of relying on a variety of specialized MRI techniques, MindGlide is able to work with basic scan types—including those not typically used for MS evaluation—unlocking insights that were previously inaccessible. The tool can analyze a scan in just 5–10 seconds.
The researchers tested MindGlide on over 14,000 images from more than 1,000 patients. Not only did it perform exceptionally well, but it also outshined two other existing AI tools—SAMSEG and WMH-SynthSeg—by a significant margin. Compared to SAMSEG, MindGlide was 60% more accurate at identifying brain lesions (also known as plaques), and 20% more accurate than WMH-SynthSeg.
What Makes MindGlide Stand Out
-
Speed and Accuracy: Rapid analysis (under 10 seconds per image) without sacrificing detail.
-
Works with Standard Scans: Even basic or older scans—like T2-weighted MRIs without FLAIR—can yield useful data.
-
Detects Brain Changes Over Time: Effective for both one-time assessments and tracking long-term progression.
-
Outperforms Experts and Existing Tools: Validated across multiple datasets with better-than-human performance.
“Using MindGlide will enable us to use existing brain images in hospital archives to better understand multiple sclerosis and how treatment affects the brain,” said Dr. Philipp Goebl, lead author from UCL’s Queen Square Institute of Neurology and Hawkes Institute.
The researchers hope this tool will help unlock millions of archived hospital images, providing a new layer of insight into how MS affects different people—and how treatments can be optimized.
“AI will unlock the untapped potential of the treasure trove of hospital information,” added Dr. Arman Eshaghi, principal investigator of the study and lead of the MS-PINPOINT group at UCL.
What’s Next?
Right now, MindGlide focuses only on the brain—it doesn’t yet analyze the spinal cord, which is also crucial for evaluating MS-related disability. Expanding its capabilities to cover the entire nervous system will be a key next step for researchers.
MindGlide was developed using an initial training dataset of over 4,000 MRI scans from nearly 3,000 patients. The model was then validated using three independent datasets totaling nearly 15,000 images. The findings were published in Nature Communications and confirm the tool’s effectiveness for both clinical trial and real-world settings.
