Multiple sclerosis (MS) is a condition in which the immune system attacks the brain and spinal cord, leading to impairments in movement, sensation, and cognition. Magnetic Resonance Imaging (MRI) markers are vital for studying MS and assessing treatment effectiveness. However, measuring these markers requires various specialized MRI scans, which limits the usefulness of routine hospital scans. Artificial intelligence (AI), which uses mathematical models to process large data sets and solve problems in ways similar to human thinking, has now led to the development of a new tool that can help evaluate treatment progress for MS patients.

The tool, named MindGlide, was developed by researchers at University College London (London, UK) to extract critical information from brain images (MRI scans) of MS patients, such as measuring brain damage, identifying subtle changes like brain shrinkage, and detecting plaques. In creating MindGlide, the team trained the AI using a dataset of 4,247 MRI scans from 2,934 MS patients across 592 MRI scanners. Through this process, MindGlide learned to identify disease markers. In a new study published in Nature Communications, the team tested MindGlide on over 14,000 MRI images from more than 1,000 MS patients. Previously, this task required expert neuro-radiologists to manually interpret complex scans over the course of years, with reports often taking weeks due to high workload. However, MindGlide was able to analyze images in just five to ten seconds and detect how different treatments affected disease progression in both clinical trials and routine care using images that were previously not analyzed.

MindGlide outperformed two other AI tools—SAMSEG, which identifies and outlines different brain regions in MRI scans, and WMH-SynthSeg, which detects and measures white matter hyperintensities important for monitoring MS. Compared to expert clinical analysis, MindGlide was 60% more accurate than SAMSEG and 20% better than WMH-SynthSeg in detecting brain lesions or plaques and assessing treatment effectiveness. The study results show that MindGlide can accurately identify and measure key brain tissues and lesions even when using limited MRI data, such as T2-weighted scans without FLAIR (a scan technique that highlights bodily fluids but also includes bright signals that can obscure plaques).

MindGlide also excelled in detecting changes in deeper brain areas, with the findings proving reliable both at a single time point and over longer periods, such as during annual patient scans. Additionally, MindGlide corroborated previous high-quality research on which treatments were most effective. The researchers hope to see MindGlide used for evaluating MS treatments in real-world settings, overcoming past limitations associated with relying solely on clinical trial data, which often didn’t represent the full diversity of MS patients. While the current version of MindGlide is limited to brain scans and does not include spinal cord imaging—a critical aspect for evaluating disability in MS—future research aims to develop a more comprehensive assessment that covers both the brain and spinal cord.

“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, first author. “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.” 

http://www.sdbailingdun.com/en/index.asp