Healthcare (Commonwealth Union) – Researchers at University College London (UCL) have created a new artificial intelligence (AI) tool designed to evaluate how effectively treatments are working for individuals with multiple sclerosis (MS).
AI systems rely on mathematical algorithms to process large datasets, enabling machines to learn and solve problems in ways that mimic human intelligence. These systems can carry out advanced tasks such as interpreting images.
The new tool, named MindGlide, can analyse brain MRI scans taken during standard MS care. It identifies critical details, including damaged brain regions, and detects subtle indicators like brain shrinkage and plaque buildup.
MS is a neurological condition where the immune system mistakenly attacks the brain and spinal cord, leading to difficulties with movement, sensation, and cognitive function. Researchers of the study pointed out that in the UK, approximately 130,000 people are affected by MS, with the condition costing the NHS over £2.9 billion annually.
MRI markers play a vital role in understanding MS and assessing the impact of treatments. However, identifying these markers often requires multiple specialised MRI scans, which many routine hospital imaging procedures are not equipped to provide.
In a recent study published in Nature Communications, scientists evaluated the performance of MindGlide by applying it to more than 14,000 brain images from over 1,000 individuals diagnosed with multiple sclerosis (MS).
Traditionally, interpreting these complex MRI scans has relied on skilled neuro-radiologists, with each case often taking weeks to process due to the heavy workload across NHS services.
For the first time, MindGlide demonstrated the ability to use AI to identify how various treatments influence disease progression in both clinical trials and everyday medical care. Impressively, it could assess images that had previously been too challenging to analyse, including standard hospital MRI scans—processing each image in just five to ten seconds.
MindGlide also outperformed two other established AI tools—SAMSEG, which maps different brain regions in MRI scans, and WMH-SynthSeg, which spots and measures bright lesions relevant to MS diagnosis and tracking. Compared with expert clinical evaluations, MindGlide proved to be 60% more accurate than SAMSEG and 20% more accurate than WMH-SynthSeg in detecting brain plaques (also known as lesions) and assessing treatment effectiveness.
The first author, Dr Philipp Goebl (UCL Queen Square Institute of Neurology and UCL Hawkes Institute), says “Using MindGlide will enable us to use existing brain images in hospital archives to better understand multiple sclerosis and how treatment affects the brain.
“We hope that the tool will unlock valuable information from millions of untapped brain images that were previously difficult or impossible to understand, immediately leading to valuable insights into multiple sclerosis for researchers and, in the near future, to better understand a patient’s condition through AI in the clinic. We hope this will be possible in the next five to 10 years.”
The study’s findings indicate that MindGlide can effectively detect and measure key brain structures and lesions, even when working with limited MRI information and scan types not typically used for this purpose—such as T2-weighted MRI without FLAIR. FLAIR usually enhances fluid contrast, but its absence can make it harder to distinguish certain abnormalities, like plaques.
MindGlide also showed strong performance in identifying changes not only in the brain’s surface but also within deeper regions.
The results were consistent and dependable across both single time points and longer-term follow-ups, such as annual patient scans.
Moreover, MindGlide was able to confirm the outcomes of previous high-quality studies regarding the most effective treatment approaches.
The research team now aims for MindGlide to be used in everyday clinical practice to assess MS therapies, helping to address the limitations of depending solely on controlled trial data, which often excludes the full spectrum of individuals affected by MS.
