Department of Neurology, Massachusetts General Hospital
Project Title: Deep learning for seizure prediction in a rat model of epilepsy
Project Summary: Nicolas worked on using iEEG recordings in an animal model of epilepsy to predict occurrence of spontaneous seizures. Mesial temporal lobe epilepsy, which is the most common form of focal epilepsy in humans, is obtained in animals with an injection of a chemoconvulsant in the hippocampus region. The goal of the project is to detect preictal phases (preceding a seizure) using feature-based machine learning approaches and deep learning/convolutional neural networks. This would allow, provided enough data is collected, to use optogenetic stimulation to change the brain networks dynamics during these preictal windows in order to preempt seizures. To that aim, Nicolas is also developing a low-cost solution to deliver light stimulation in deep brain structures, based on μLEDs.