AI algorithm helps cure epilepsy
FCDs are regions of the brain that develop abnormally and often cause drug-resistant epilepsy. It is usually treated surgically, however identifying lesions from MRI is an ongoing challenge for clinicians, as MRI scans in FCD can look normal.
To develop the algorithm, the team quantified cortical features from MRI scans, such as how thick or folded the cortical/brain surface is, and used about 300,000 locations across the brain.
The researchers then trained the algorithm on examples that were labeled by radiologists as healthy brains or with FCD – depending on the patterns and their features.
The findings, published in Brain, show that overall, the algorithm was able to detect FCD in 67% of cases in the cohort (538 participants).
Previously, 178 of the participants were considered MRI negative, meaning radiologists were unable to find an abnormality – but the MELD algorithm was able to identify FCD in 63% of these cases. .
This is especially important, because if doctors can find an abnormality during brain scans, surgically removing it could provide a cure.
“Our emphasis is on creating an AI algorithm that is interpretable and can help doctors make decisions,” said first co-author Mathilde Ripart (UCL Great Ormond Street Institute for Children’s Health). essential part of that process.”
Senior co-author Dr Konrad Wagstyl (UCL Queen Square Institute of Neurology) added: “This algorithm could help find more potential lesions in children and adults with epilepsy, co-authoring the study. time allows more epilepsy patients to be considered for brain surgery which can cure epilepsy and improve their cognitive development About 440 children a year can benefit from epilepsy surgery in the UK. ”
In the UK, around 600,000 people are affected. While drug treatments are available for the majority of people with epilepsy, 20-30% do not respond to medication.
In children who have had surgery to control epilepsy, FCD is the most common cause, and in adults it is the third most common cause.
In addition, for patients with epilepsy whose brain abnormalities cannot be found on MRI, FCD is the most common cause.
Co-first author Dr. Hannah Spitzer (Helmholtz Munich) said: “Our algorithm automatically learns to detect lesions from thousands of patients’ MRI scans. It can detect it reliably. trust lesions of different types, shapes, and sizes, and even many of those lesions previously missed by radiologists.”
Co-lead author Dr Sophie Adler (UCL’s Great Ormond Street Institute for Children’s Health) added: “We hope that this technology will help identify abnormalities that cause epilepsy that are currently overlooked. . Ultimately, it may allow more people with epilepsy to have potentially curable brain surgery.”
This study of FCD detection uses the largest group of MRI resonances of FCD to date, which means it can detect all types of FCD.
The MELD FCD classifier can run on any patient with suspected FCD who is over 3 years old and has had an MRI scan.
The MELD project is supported by the Rosetrees Trust.