Scientists at Newcastle University have used brain scans from patients with the most common type of epilepsy - temporal lobe epilepsy - and computer modelling techniques to look at the brain as an example of a computer network.

By simulating brain activity within each patient-specific network, they successfully identified regions that were more prone to seizures.

The research is published today in PLOS Computational Biology, and is believed to be the first study to combine computational modelling of brain dynamics with patient-specific MRI data from individuals with temporal lobe epilepsy.  

Around 1% of the UK population suffers from epilepsy and, in many cases, it is an extremely debilitating illness. Currently, anti-convulsant drugs are the main treatment but these are not always effective. In these cases, surgical removal of the parts of the brain indicated by EEG readings to be the source of the seizure is carried out. However, in about 30% of cases, surgery does not result in preventing seizures. 

Identifying the most seizure prone areas

The research team, based in the School of Computing Science at Newcastle University, simulated surgery by disconnecting sections of the network that corresponded to the parts of the brain most commonly removed. They also ran individual patient simulations, removing the most seizure-prone parts of the model for each person. By mimicking seizures before and after ‘surgery’, they found that patient-specific surgery showed, in every case, a significant improvement compared to removal of the regions most commonly taken out.

Dr Peter Taylor, who co-led the study, explained: "This research may help to explain why surgery is so often unsuccessful, as this work predicts that the areas most commonly removed in surgery are not always involved in starting and spreading seizures.

“It also takes us a step further towards rectifying the problem, as identifying the most seizure prone areas on an individual basis has the potential to show when the usual surgery procedures may not work for a patient.”

Research lead Frances Hutchings added: “Removal of brain tissue is often the final option for treatment of temporal lobe epilepsy but we know that it is not always effective. It’s early days and there is more work to be done, but this model could assist surgeons in targeting surgical procedures more effectively and help people with epilepsy lead a more normal life.”

In future, the team intend to check the model’s predictions against patient-specific surgical outcomes. Professor Marcus Kaiser, Professor of Neuroinformatics at Newcastle University, said: “The next steps are to compare the computationally predicted outcomes with the actual surgery outcomes in individual patients and to investigate how alternative surgery targets can be included in the future treatment.”

Reference: Hutchings F, Han CE, Keller S, Weber B, Taylor PN, Kaiser M. Predicting Surgery Targets in Temporal Lobe Epilepsy through Structural Connectome Based Simulations. 

PLOS Computational Biology, December 2015 http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1004642