Paper Title
Epileptic Seizure Prediction For Multi-Channel Scalp Electroencephalogram With 3d-Convolutional Neural Networks

Abstract
Traditional method of epileptic seizure prediction could not avoid the process of manually selecting the features. Recently, the development of deep learning technology has provided a new direction. This paper introduces3D-convolutional neural networks (CNN) and pre-processing methods to predict epileptic seizures using scalp electroencephalogram (sEEG) signals. In the proposed method, Power spectral data were mapped as a 2D image to preserve characteristics of sEEG signals at each electrode. The proposed neural network extracts spatial and temporal features from the pre-processed 3D structure. Then, the comparison with the conventional method has been done, which shows that the approach presented in this paper is better. The proposed approach achieves sensitivity of 90.8%, 91.10% and false prediction rate of 0.134/h, 0.119/h on Children’s Hospital of Boston-MIT sEEG dataset and Seoul National University Hospital sEEG dataset, respectively. All of those indexes reveal the high performance and stability of the approach for the epileptic seizure prediction. Keywords - Epileptic seizure, convolutional neural networks, deep learning, seizure prediction, sEEG