A Comparison of Deep Neural Networks for Seizure Detection in EEG Signals

Abstract

This paper aims to apply machine learning techniques to an automated epileptic seizure detection using EEG signals to help neurologists in a time-consuming diagnostic process. We employ two approaches based on convolution neural networks (CNNs) and artificial neural networks (ANNs) to provide a probability of seizure occurrence in a windowed EEG recording of 18 channels. In order to extract relevant features based on time, frequency, and time-frequency domains for these networks, we consider an improvement of the Bayesian error rate from a baseline. Features of which the improvement rates are higher than the significant level are considered. These dominant features extracted from all EEG channels are concatenated as the input for ANN with 7 hidden layers, while the input of CNN is taken as raw multi-channel EEG signals. Using multi-concept of deep CNN in image processing, we exploit 2D-filter decomposition to handle the signal in spatial and temporal domains. Our experiments based on CHB-MIT Scalp EEG Database showed that both ANN and CNN were able to perform with the overall accuracy of up to 99.07% and F1-score of up to 77.04%. ANN with dominant features is more capable of detecting seizure events than CNN whereas CNN requiring no feature extraction is slightly better than ANN in classification accuracy.

An improved method for recognizing pediatric epileptic seizures based on advanced learning and moving window technique

“…Although approximate energy has been used as a feature [17], in the present study using the test sample method energy has given the lowest accuracy across all the classifiers. When compared with prior works, the proposed approach performed well except for few works [8,17,32]. Justification can be given as all the above-mentioned works used the patient-specific approach whereas for the present work a generalized seizure detection framework has been maintained for each classifier so that seizures pertaining to any type or location can be detected with maximum sensitivity.…”

“…Table 9 represents the comparative study of the existing work with previous works based on the same dataset. [8] used 18 channels out of the 23 channels for seizure detection for the same data set. But in the proposed method all the channels are used to propose a channel independent method.…”

“…Most of the work suggested for epilepsy detection revolves around epilepsy instead of pediatric epilepsy. Detection of pediatric epileptic seizure is difficult in nature and studies tend to concentrate on patient specific approach [8,17,32,34,37,41]. Few works focused on channel specific approach [8,39] instead of generalized approach.…”

“…Detection of pediatric epileptic seizure is difficult in nature and studies tend to concentrate on patient specific approach [8,17,32,34,37,41]. Few works focused on channel specific approach [8,39] instead of generalized approach. In other suggested literatures, the accuracy of detection is not as promising [34,41].…”

Epileptic Seizures Detection Using Deep Learning Techniques: A Review

“…In [ 80 ], 1D-CNN was used for feature extraction procedure. The researchers in [ 81 ] used 1D-CNN for other work. They used a CHB-MIT dataset, and the signals from each channel were segmented into 4 s intervals; overlapping segments were also accepted to increase the data and accuracy.…”