A method to delineate epileptic brain areas remotely

Brain surgery is a key treatment option for patients with drug-resistant epilepsy. During the surgery, it is important to delineate accurately the brain area that causes epilepsy. To analyze electrical recordings, the Sarnthein and Indiveri research groups have developed and tested a special kind of computer hardware inspired by brain function that can detect epileptic activity during surgery.

Using neuromorphic technology developed in the research group of Giacomo Indiveri at the Institute of Neuroinformatics of the University of Zurich and ETH Zurich, the team of Johannes Sarnthein at the University Hospital Zurich developed a system able to analyze electric brain signals recorded in real-time during brain surgery. Using a spiking neural network, they detect epileptiform patterns that are typical markers of the epileptogenic brain area.

They first tuned their system on a public dataset from Zurich that they had previously analyzed with standard computers. They achieved the same prediction of surgical success for all 22 patients.

They then validated their fully automated system in a real-time setting. Electric brain signals recorded during surgery in Utrecht were compressed and streamed for remote processing in Zurich. This demonstrates real-time detection of epileptiform patterns, which could advise surgeons during epilepsy surgery. The remote processing illustrates how the system could be used ‘as-a-service’ in clinical practice.

Costa, F., Schaft, E.V., Huiskamp, G. et al. Robust compression and detection ofepileptiform patterns in ECoG using a realtime spiking neural network hardware framework. Nature Communications (2024) 15:3255. https://doi.org/10.1038/s41467-024-47495-y

Additional information:

Image: iStock