AI can detect minute brain abnormalities to treat epilepsy

STORY: :: King’s College London/UCL

This MRI scan of a human brain shows in red an area that could be a potential lesion, and a possible cause of epilepsy.

And a new artificial intelligence tool developed in the UK can help spot the sometimes minute and difficult-to-pinpoint abnormalities that can cause epilepsy, and prompt methods of treatment.

:: ON AI

:: Dr Konrad Wagstyl, King’s College London

“The crucial thing is, if you can find the brain abnormality, you can offer surgery which might cure the seizures. And what our AI tool can do is find two thirds of those abnormalities that doctors might miss.”

Dr Konrad Wagstyl is a Senior Lecturer at the School of Biomedical Engineering & Imaging Sciences, King’s College London.

He’s worked to develop this tool, called MELD Graph. It came together with the help of 23 hospitals around the world that shared their records of brain abnormalities with researchers.

That helped them construct a dataset on which to train the AI algorithm on what to look for.

“It can take a new scan, point out where it thinks the abnormality is on that scan, and also really importantly, how confident it is, nine out of ten, ten out of ten that it’s a brain abnormality. But also what it’s using to make that decision. Is the brain a bit too thick in this area? Is it a bit brighter than it’s expecting? Then a radiologist can go back to that report and have a look at the scan and say, ‘yes, I think this is causing the seizures. Maybe we can discuss surgery now’.”

Epilepsy affects around 1 out of every 100 people globally, with 1 in 5 epilepsy cases being caused by a structural abnormality in the brain.

The variations can be minuscule.

“…an MRI scan is maybe 10 million pixels and they’re looking for something that’s maybe 100 pixels at its smallest. And that can be very subtle. So for context, if you took, an example might be a novel, a 200 page novel has 100,000 words in it and they’re looking for one word that’s slightly the wrong size or slightly the wrong font. So you could see how having an AI to point out potential areas can really speed up that process and also pick up things they might have missed.”

While the tool is not yet clinically available, the team has made the software open-source and is training clinicians and researchers worldwide on its use.