Deep, personalized epileptic seizure detection

There is an urgent need for objective epileptic seizure monitoring in the home environment.  The only reliable way of monitoring seizures is by measuring brain activity, which is possible by means of the electroencephalogram (EEG).  Miniaturized EEG devices with different electrode configurations that could be worn at home became available, but the challenge remains how to automatically detect the presence of seizures.  Having demonstrated previously the potential of deep learning in the context of automated EEG-based sleep analysis, the aim of this project is three-fold.  First, we aim to demonstrate that appropriate deep learning approaches trained on various databases can lead to reliable automated seizure detection, also in the home environment.  Second, we propose a new way of extracting representations from EEG data.   This approach has particular advantages as it should reduce the need for large amounts of labeled data and it has the potential of being more robust to artifacts.  Third, as brain waves dramatically vary between subjects, we will personalize the deep learning approaches in supervised and unsupervised ways in order to optimise the performance further. Also here, the novel representation will be advantageous.