Intelligent signal classifier for brain epileptic EEG based on decision tree, multilayer perceptron and over-sampling approach

Abstract

Epilepsy is a chronic neurological disease induced by abnormal electrical discharges of brain which tends to irregular seizures. The seizures may cause the patients to lose consciousness and the patients couldn’t control their muscles. Epilepsy even possibly endangers one’s life. Electroencephalogram (EEG) is a common tool used in the clinical diagnosis and analytics of epilepsy. However, the visual examination of EEG is time-consuming and the diagnostic result is also easily influenced by the viewer’s subjective judgement. Therefore, the purpose of this study is to construct an automatic classifier, which could be helpful to analyze, for the epileptic EEG signals. The EEG recordings of patients with intractable epilepsy, which are collected by Boston Children’s Hospital, are used in this study. The features of EEG signals in time and frequency domains are collected from results of the Fast Fourier Transform. The Synthetic Minority Oversampling Technique (SMOTE) is used to solve the data imbalance problem. Four machine learning algorithms including C4.5, Classification and Regression Tree (CART), Chi-Square Automatic Interaction Detector (CHAID) and Multilayer Perceptron (MLP) are used to classify the data. As a result, the accuracy rate of the proposed classifier is 99.48%. It might be a clinical assistant tool for doctors to make a more reliable and objective diagnosis.