Using AI to identify brain abnormalities
Emerging Leader Fellowship
£144,468
Dr Jonathan Horsley
Newcastle University
Focal epilepsy arises when seizures originate in a specific area of the brain, and MRI scans play a central role in guiding treatment decisions, including surgery. However, in many people with focal epilepsy, MRI scans appear normal even when subtle abnormalities are present. When these hidden abnormalities are not detected, patients may miss the opportunity for surgery or other more effective treatments.
This project will develop computer models that learn what a healthy brain looks like using tens of thousands of MRI scans from people without epilepsy. These models will then compare scans from people with epilepsy against this reference to identify subtle abnormalities that may represent seizure-causing tissue. The work will first test whether these models can reliably detect abnormalities across different MRI types. It will then evaluate whether the identified regions correspond to tissue removed during epilepsy surgery and whether their removal is linked to better seizure outcomes in over 1,000 surgical cases. Finally, the project will investigate whether different causes of epilepsy produce distinct imaging “signatures” by comparing MRI findings with microscopic analysis of surgical brain tissue.
This research could improve the detection of subtle brain abnormalities that are currently missed on standard MRI scans, leading to more accurate diagnosis and treatment planning. In the short-to-medium term, it may help better identify candidates for surgery and other targeted treatments. In the long term, it could improve surgical outcomes and support the development of more precise, personalised therapies for people with epilepsy, increasing the chance of seizure freedom.
