£148,278 over 36 months
Awarded in 2022


Multi-centre Epilepsy Lesion Detection Project: a collaborative cohort for the analysis of focal epilepsies


Dr Sophie Adler


Dr Konrad Wagstyl (Wellcome Centre for Human Neuroimaging); Prof. Torsten Baldeweg (UCL, Institute of Child Health); Prof. Helen Cross (UCL, Institute of Child Health); Prof. John Duncan (UCL, Institute of Neurology); Dr Juan Eugenio Iglesias (UCL, Medical Physics & Biomedical Engineering)


UCL (Great Ormond Street Institute of Child Health)

Structural brain abnormalities that cause complicated epilepsy that doesn’t respond to medications are often subtle and missed on MRI scans. We need to capitalise on developments in Artificial Intelligence to develop tools to automatically find epilepsy-causing abnormalities on MRI and predict which patients will be seizure free after surgery. Our Multi-centre Epilepsy Lesion Detection (MELD) Project, by collecting anonymous MRI scans from 100s of patients worldwide, will create these tools and make them available to the epilepsy community.
Dr Sophie Adler


In many people with epilepsy, seizures are caused by structural abnormalities, such as areas of the brain that have developed abnormally or certain types of tumours. These structural abnormalities often cause drug-resistant epilepsy, where drugs are unable to prevent seizures. For these people, surgery to remove the structural abnormality can stop seizures. However, the abnormalities can be hard to find and completely remove, and surgery is only successful in 6 out of every 10 patients. Dr Adler’s overall aim is to improve these outcomes by developing novel Artificial Intelligence (AI) algorithms that will automatically find these subtle abnormalities on patients’ MRI scans and help neurosurgeons to plan operations that will completely remove them.


Dr Adler’s Multi-centre Epilepsy Lesion Detection (MELD) project will create the largest collection of anonymised MRI data from patients with epilepsy caused by structural abnormalities from hospitals world-wide. The team will use this unique dataset to create atlases of where these structural abnormalities occur in the brain, helping us to understand why they cause epilepsy. From this, the team will train AI algorithms to find where on the MRI the abnormalities are and to diagnose the cause of the epilepsy. They will then be able to develop an algorithm to predict how patients will respond to surgery, which neurosurgeons could use to simulate and plan surgical removal of the abnormal brain area whilst keeping important brain areas intact – for example, those used for language, movement and vision.


The findings of this research could be implemented in pre-surgical consultations to inform patients and their families of the likelihood of surgical success. Importantly, the AI algorithms developed will be made freely available to use, helping to improve the diagnosis and planning of surgeries for patients with epilepsy in the UK and across the globe. Through the MELD project, Dr Adler will develop tools for MRI diagnosis and surgical planning of patients with epilepsy caused by structural abnormalities. These will be used by academic neurologists, neuroradiologists and neurosurgeons who will be able to use them in patient pathways through clinical trials. A previous algorithm developed by the team is already part of a clinical trial at Great Ormond Street Hospital, and is being used at the National Hospital for Neurology and Neurosurgery and as a research tool internationally.