Artificial intelligence (AI) could help detect brain abnormalities linked to epilepsy, leading to earlier diagnosis, researchers have revealed.
The software, named Meld Graph, was developed by scientists at King’s College London and University College London (UCL). It aims to assist radiologists in identifying abnormalities that might otherwise be missed.
Meld Graph was trained to detect focal cortical dysplasia (FCD), malformations caused by neurons not developing properly in certain brain regions. These lesions are a common cause of drug-resistant epilepsy, affecting thousands of patients.
In the UK, 630,000 people live with epilepsy, with around one in five experiencing epilepsy caused by brain lesions. While these can be treated surgically, FCDs are often subtle and difficult to identify, experts say.
Dr Konrad Wagstyl, of King’s College London, explained: “Radiologists are overwhelmed with images. AI tools like Meld Graph can support them, improving NHS efficiency and accelerating treatment for patients.”
For their study, published in JAMA Neurology, researchers analysed MRI data from 1,185 patients across 23 epilepsy centres worldwide, including 703 with FCD. Half of the participants were children.
Meld Graph successfully detected 64% of brain malformations, prompting optimism about its clinical potential. Dr Luca Palma, from Bambino Gesu Children’s Hospital, highlighted its success in diagnosing a 12-year-old boy whose seizures had persisted despite multiple medications.
Although not yet clinically available, Meld Graph has been released as open-source software. The team is running workshops to train clinicians worldwide, including in the UK, Chile, India, and France.
Great Ormond Street Hospital is among the institutions adopting the tool. Professor Helen Cross, consultant epileptologist, stressed its importance, saying: “Many children experience years of seizures before we find a lesion. This technology could change that.”
