EEG Signal Preprocessing Study Based On Wavelet Transform And Independent Component Analysis

Electroencephalogram(EEG) is the spontaneous electrical activity recording of brain cells from scalp.EEG data is recorded by the electrodes sticked on the scalp.So,the data are easily influenced by some other components which will interfere us to analyse the data.It is important to remove artifacats from EEG recording for further study. As one of the modern signal-processing tools for non-stationary signal,Wavelet Transform(WT) with multi-distinguish identity can focus on every details of the signal,and it's band-pass characteristic can help us pick-up or remove some of the narrowband signals from the original signals expediently; Independent Component Analysis(ICA) is a new signal-processing method based on blind sources separation,which is useful in processing mulriple channels EEG signals.This paper focus on several topics described as following:Contrastively study the whole bag of tricks for removing artifacts by respectively experiments,aim at several familiar EEG artifacts(such as:power noise,electrocardiogram(ECG),electrooculogram(EOG)) recorded at different conditions,and find out the best resultful method.The study results show that Wavelet Transform(WT) and Independent Component Analysis(ICA) methods are provided with universality applicability in EEG signal preprocessing..Especially,when the observation signal channels number less than the independent original signal channels number and without reference signal recorded at the same time,usually can use WT method to remove artifacts.And when the observation signal channels number more than or equal the independent original signal channels number,ICA method is the best choice.Besides,if have the reference signal recorded at the same time,ICA can separate the artifacts from the EEG signal more effectively.Based on the above study results,this paper put forward a new method which unite WT with ICA,and apply it in the epilepsy EEG signal preprocessing.The experiment result testify that this method picked-up the epilepsy characteristic waves usefully,and make good preparations for further study epilepsy waves.This method can be consider as a universal,impactful and steady EEG signal preprocessing method.It can be applied in removing several artifacts at the same time,and buildup the characteristic signal needful.