Performance Analysis of the Classifier in the Classification of Normal-Sleep and Seizure from EEG Signal

Abstract

An electroencephalogram is a medical method that employs electrical signals to analyze brain activity. The (EEG) signal is commonly measured using Scalp electrodes, which is very useful in identifying a patient's brain status and epilepsy as well as supplementing CT scan measurements. EEG signals indirectly reveal the state of the brain. In this paper the performance of the classifiers are analyzed to detect Normal sleep and Seizure EEG signals. Features are extracted using six statistical features such as Mean, Variance, Skewness, Kurtosis, sample entropy, Pearson Correlation coefficient. The Detrend Fluctuation Analysis, Detrend Fluctuation Analysis Expectation Maximization, Detrend Fluctuation Analysis Firefly, Detrend Fluctuation Analysis with Gaussian Mixture Model, Detrend with Bayesian Linear Discriminant Classifiers are employed to detect the Normal sleep and Seizure from EEG signal. The hybrid classifier Detrend Fluctuation Analysis with EM achieved the highest accuracy of 98.96% for Seizure EEG signal and an accuracy of 97.66% using Detrend Fluctuation Analysis classifier for normal sleep EEG signal.