Application Of Independent Component Analysis And Wavelet Threshold In De-noising Of Epileptic EEG Signals

Electroencephalography(EEG)is the recording of electrical activities of human brain that carries a large amount of information about physiology and pathology.It has been verified that EEG is a valuable tool for the study of the human brain function,behavior consciousness,rehabilitation engineering etc.,especially for the diagnosis and treatment of epilepsy as well as Alzheimer's diseases and so on.However,the recorded EEG signals are highly susceptible to the non-brain tissues and surrounding environments,which results in a variety of artifacts.These artifacts often cover up some useful but weak information in EEGs,and hence will influence the further analysis of EEG signals.Therefore,there has been an increasing interest in the study of EEGs denoising in recent years.Although a lot of progresses have been made,there still exist some problems,including noise identification,selection of reference noise etc.Under such situation,this paper first proposes a novel denoising method for EEG signals,which is based on the independent component analysis(ICA)and wavelet threshold.Then combining existing feature extraction methods and extreme learning machine(ELM),epileptic seizure detection has been completed automatically to verify the performance of the method.The main contents are presented as follows:Chapter 1 systematically introduces the background and development of the denoising methods for EEG signals.Chapter 2 introduces the theory of ICA,including the mathematical preliminary,information theory foundation and basic models.Then the principle of related algorithms are deduced.Chapter 3 describes the basic characteristics of EEG signals and proposes a novel denoising methods,which is based on ICA and wavelet threshold.Chapter 4 introduces the basic procedure of automatic seizure detection using EEGs,and then the performance of our proposed method is verified by numerical simulations.