Study On The Removal Of The Ocular Artifacts From EEG Data

The amplitude of electroencephalogram (EEG) is very small,with intensity aroundmicrovolt.Artifacts such as electrocardiogram (ECG),electromyography (EMG) andelectro-oculogram (EOG) often contaminate the EEG signal.It is very difficult to readand analyse the EEG signal in these contaminations,especially for automatic analysisand diagnoses.How to reduce or reject these contaminations from EEG has become avery important issue.Among these contaminating sources,EOG is the major one.Theydistort the electric field distribution of actual brain activities over the scalp,especiallysurrounding the eyes.The study of eye artifact is not only a theoretical concern ofresearch,but also an important issue for EEG application in clinic.Almost all usersmust take into account the effect of ocular activities on EEG,and the method of EOGremoval should be reported in publications.In recent 20 years,with the availability ofdigital EEG,it has become a desirable procedure to correct artifacts automatically bycomputer.In this article,the following EOG removal methods,which are based on scalp EEGdata and equivalent distributed source theory,are presented and discussed.1.An ICASC method is presented to remove EOG artifact,which is based onIndependent Component Analysis (ICA) and Space Correlation.Firstly,ICA isutilized to decompose the scalp EEG contaminated by EOG.Second,the correlationcoefficients between the spatial distributions of the decomposed independentcomponents and the EOG are computed to identify the artifact components.Finallythe artifact components are removed and the corrected EEG is achieved.Thevalidity of this method was poved by actual EEG data.2.PCAF and PCAR are presented to remove the EOG artifact.The PCAF method isbased on Principal Component Analysis (PCA) and frequency- domain analysis.Firstly among all of the decomposed components,the EOG principal components are filtered according to the EOG frequency range.Then the filtered components areremoved.The PCAR method based on PCA and the Regression Algorithm.TheEOG components,decomposed by PCA,were used to substitute the EOG signal inthe Regression Algorithm to get the attenuation factor.The space distribution ofEOG components are revised by the attenuation factor.Then the EOG componentsare adjusted.The adjusted EOG components are utilized to remove the EOGartifacts according to the traditional PCA method.The validity of the two methodsare tested by actual EEG data and simulated data.3.A CAscaded Spatio-Temporal processing procedure (CAST) is presented to removeEOG artifact.Firstly,the discrete equivalent distributed source on the corticalsurface is reconstructed from the contaminated scalp recordings by linear minimumnorm estimation.Secondly,PCA method is utilized to decompose the equivalentsource and then the EOG component is indentified and removed.Finally,artifactfree scalp EEG is reconstructed from the equivalent distributed source where EOGcomponents have been removed.The effectiveness of CAST is confirmed by theapplication to real scalp data and comparison with other methods.4.An ACAST method is presented to remove EOG artifact automatically.The ACASTis consisted of the CAST method and an EOG artifact recognition algorithm basedon the wave shape and amplitude of EOG signal.And the validity of this method isevaluated by the experimental data of Inhibition of Return (IOR).5.An electrode cap is presented for acquiring the EEG signal.This cap can fasten theelectrodes tightly to the scalp and avoid the electrodes flipping away.And theelectrodes in this cap can move freely according to the need of acquisition.This capcan also measure the position of electrodes and the shape of subject head.