AI tools can track the effectiveness of multiple sclerosis treatment

UCL researchers have developed a new artificial intelligence (AI) tool that can help explain and evaluate good treatment in patients with multiple sclerosis (MS).

AI uses mathematical models to train computers using large amounts of data to learn and solve problems in ways that look like humans, including performing complex tasks such as image recognition.

This tool is called Mindglidekey information can be extracted from brain images (MRI scans) obtained during care of MS patients, such as measuring damaged areas of the brain and highlighting subtle changes such as brain contractions and plaques.

MS is a condition where the immune system attacks the brain and spinal cord. This can lead to a person’s movement, feeling or thinking about problems. In the UK, 130,000 people live with MS, which costs more than £2.9 billion a year.

Magnetic resonance imaging (MRI) labeling is essential for the study and testing of the processing of MS. However, measuring these markers requires different types of specialized MRI scans, limiting the effectiveness of many routine hospital scans.

As part of a new study, published in Natural communication, The researchers tested it Mindglide More than 14,000 images from over 1,000 MS patients.

This task previously required expert neuroradiologists to manually explain the years of complex scans – reporting turnover time for these images is usually due to the workload of the NHS.

But this is the first time Mindglide Ability to use previously unanalyzed images and conventional MRI scans can successfully use AI to detect how different treatments affect disease progression in clinical trials and routine care. Each image takes only five to 10 seconds.

Mindglide It is also better than the other two AI tools (tools for identifying and overviewing and outlining different parts of the brain in MRI scans) and WMH-SynthSeg, a tool that detects and measures seen and measured in certain MRI scans, for the diagnosis and monitoring of MS (such as MS), and performs better with the other two AI tools.

Mindglide 60% better than SAMSEG and 20% better than WMH-Synthseg, it is used to locate brain abnormalities called plaques (or lesions) or monitor treatment effects.

First author, Dr. Philip Goble (UCL Queen Square Neurology Institute and UCL Hawkes Institute), said: “Use Mindglide It will allow us to use existing brain images in hospital archives to better understand multiple sclerosis and how treatments affect the brain.

“We hope that the tool will unlock valuable information from millions of unexplored brain images that were previously difficult or incomprehensible, which immediately leads to valuable insights into the researchers’ multiple sclerosis, which will better understand the patient’s condition in the near future through AI in the clinic. We hope this may be possible in the next five to ten years.”

The results of the study show that it can be used Mindglide Even using limited MRI data and a single type of scan type that is not usually used for this purpose, such as T2-weighted MRI (a T2-weighted MRI without a proper amount) even if important brain tissue and lesions are accurately identified and measured (a scan that highlights the fluid in the body but still contains bright signals) – making it difficult to see Plaques).

and better performance in detecting changes in the outer layer of the brain Mindglide It also performs well in deeper brain areas.

These findings were valid and reliable at one time point and longer (i.e., annual scans in which patients participated).

also, Mindglide Able to confirm previous high-quality studies on which treatments are most effective.

Researchers now hope Mindglide Can be used to evaluate real-world MS processing, overcoming the limitations of previously relying solely on high-quality clinical trial data that typically do not capture the full diversity of MS.

“We have not previously analyzed most clinical brain images due to their lower quality. AI will be due to its lower quality. AI will release the potential of unexplored hospital information, which can give you an unexplored potential in your mind, affect your unverified influence and be able to identify effects,” said UCL Hawkes Institute, lead researcher for the project and lead researcher for the MS-Pinpoint group.

Research limitations

Current implementation Mindglide Confined to brain scans and not spinal cord imaging is important for evaluating disability in patients with MS. Future research will require a more comprehensive assessment of the entire nervous system to cover the brain and spinal cord.

develop Mindglide

Mindglide It is a deep learning (AI) model developed by UCL researchers to evaluate the brain’s MRI image and determine the damage and changes caused by MS. In development Mindglide The scientists used an initial dataset of 4,247 brain MRI scans from 2,934 MS patients in 592 MRI scans. In the process Mindglide Train yourself to identify markers of disease. This new study is for verification Mindglidethree separate databases for 14,952 images from 1,001 patients.