Epileptic Seizure Prediction: A Multi-Scale Convolutional Neural Network Approach

This study aims to develop an efficient and reliable epileptic seizure prediction method using intracranial EEG (iEEG) data. The prediction procedure should yield accurate results in a fast enough fashion to alert patients of impending seizures. We first introduce an efficient pre-processing methodology for reducing the data size and converting the time-series iEEG data into an image-like format. A multi-scale convolutional neural network architecture is then used for iEEG feature extraction and classification. Seizure prediction results show that our algorithm notably outperforms existing methods by achieving an average sensitivity of 87.85% and area under the ROC curve of 0.84.