BackgroundElectroencephalography (EEG) is widely used in the diagnosis of epilepsy, but it relies on a person having a seizure whilst being monitored. This can lead to delays in diagnosis and treatment, unnecessary anxiety and reduced quality of life.To try and address this problem, Professor John Terry, at the University of Exeter, has been working alongside neurologists in London to develop computer models that can detect ‘hidden’ information within brain networks, in short resting EEG recordings (during which no seizure has taken place), and accurately identify whether or not a person has generalised epilepsy.The studyIn 2012, Professor Terry and his team were awarded £139,595 to refine their models and find out whether they could:
- differentiate more accurately between people with and without generalised epilepsy.
- distinguish, based on differences in brain network properties, between people with focal and generalised epilepsy.
- identify, based the activity of neuronal networks, people who have responded to antiepileptic drug (AED) treatment and those who have not.
If successful, the models would have real potential to enhance the diagnosis of epilepsy, and perhaps even allow neurologists to predict a) whether or not a person would respond to an AED, and b) what the best AED treatment for a person might be, thus reducing the time to optimal therapy.This grant has now come to an end and the final report has been submitted.ResultsWith regards to aims one and two above, the findings are extremely encouraging. The researchers now have a model that can distinguish with considerable accuracy, from resting EEG recordings, groups of people with generalised epilepsy from healthy controls (who do not have epilepsy). Professor Terry reports that in a test of 30 people with epilepsy and 38 without epilepsy, the model had a misdiagnosis rate of less than 5%. This is highly significant, and the team has already been exploring how they might incorporate the model into a clinical device. On the advice of feedback from commercial companies, they intend to develop a working prototype in the near future.In terms of differentiating between people with focal and generalised epilepsy, the preliminary data obtained through this grant suggest that this should indeed be possible using computer network modelling. Professor Terry was awarded another ERUK project grant in 2015 to progress these findings and we look forward to hearing the outcome.For a number of reasons, mainly an unexpected lack of viable resting EEG data from before and after AED treatment, it wasn’t possible to make a lot of progress on aim three. However, there has been significant follow-on funding generated from this grant, particularly from EPSRC, which will enable the team to carry out this research.SignificanceThis grant has further highlighted the potential of computer models as clinical tools for the diagnosis and management of epilepsy. There is a lot more work to be done to establish their full potential, but the funding is in place and the preliminary evidence is encouraging. We are very excited about this work, as it stands to make a real difference to people’s treatment journeys and quality of life.Click here to view our research portfolio.