Epileptic Seizure Prediction With Multi-View Convolutional Neural Networks

Chien-Liang Liu,Bin Xiao,Wen-Hoar Hsaio,Vincent S Tseng

doi:10.1109/access.2019.2955285

Abstract

The unpredictability of seizures is often considered by patients to be the most problematic aspect of epilepsy, so this work aims to develop an accurate epilepsy seizure predictor, making it possible to enable devices to warn patients of impeding seizures. To develop a model for seizure prediction, most studies relied on Electroencephalograms (EEGs) to capture physiological measurements of epilepsy. This work uses the two domains of EEGs, including frequency domain and time domain, to provide two different views for the same data source. Subsequently, this work proposes a multi-view convolutional neural network framework to predict the occurrence of epilepsy seizures with the goal of acquiring a shared representation of time-domain and frequency-domain features. By conducting experiments on Kaggle data set, we demonstrated that the proposed method outperforms all methods listed in the Kaggle leader board. Additionally, our proposed model achieves average area under the curve (AUCs) of 0.82 and 0.89 on two subjects of CHB-MIT scalp EEG data set. This work serves as an effective paradigm for applying deep learning approaches to the crucial topic of risk prediction in health domains.

Highlights

50 million individuals worldwide have been diagnosed as having epilepsy [8], but existing treatments such as surgery and anticonvulsants pose severe side effects to patients
EXPERIMENTAL RESULTS To evaluate the effectiveness of the proposed multi-view convolutional neural network (CNN), we conduct experiments on two data sets, and the experimental results show that the proposed method outperforms all methods listed in the Kaggle leader board
WORK This work proposes a multi-view CNN framework to predict the occurrences of seizures

Summary

Introduction

50 million individuals worldwide have been diagnosed as having epilepsy [8], but existing treatments such as surgery and anticonvulsants pose severe side effects to patients. Epilepsy, which is characterized by unpredictable seizures, adversely affects patient mental health, often resulting in anxiety, depression, or cognitive impairment [26]. An accurate prediction model that provides an alert prior to the occurrence of seizure can considerably improve patient quality of life. Most studies have relied on electroencephalograms (EEGs) [5], [16], [39], which are recordings of the electrical potential of the brain, to capture physiological measurements of epilepsy. Note that the EEG signals could be presented in frequency domain and time domain. The two domains provide two different views for the same data source, in which the time domain measures how the

Objectives

Results

Discussion

Conclusion