Application Of Independent Component Analysis In EEG Signals Processing

In the EEG signals processing,the EEG signals colleted from electrodes on scalp result from the electricity activity of neuron in the different region of brain,including of bioelectricity (Electrocardiograph ECG,Electromyogram EMG, Electro-Oculogram EOG)from other organs and tissues,and noise arose from the outer factors. The purpose of EEG signal processing is to extract the hidden or weak paterns that probably have some physiological and/or psycho-physiological significance from EEG signals in sophisticated noise background and then to apply them to the research on clinical medicine or cognitive science.However, abandoning the EEG signals contaminating by the noise can lose a lot of data definitively,which is difficultly accepted by the clinic . So,in the EEG signal processing, it is a difficult problem how to acquire the useful information reflecting the brain's activity and state from the original EEG Signals. The traditional approaches for EEG signal processing are mostly on the time or frequency analysis. But due to the strong randomness and nonstationarity of EEG the results obtained from those traditional methods are not very satisfying.Independent component analysis(ICA) is a new signal separating technique in the field of statistical signal processing.IC A is featured by decomposing the observed record into mutually independent components.The new method is now being noticed by researchers in biomedical engineering.In this thesis the theory of independent component analysis as well as its application in EEG signal processing are studied.In the paper,the model of ICA,its contrast functions,and algorithms are analyzed.The contrast functions of ICA mainly include the measurement of nongaussianity,minimization of mutual information,infomax method,and maximum likelihood estimation.The algorithm based on high order cumulant,and the algorithmbased on neural network are the generally used ICA algorithms.Fixed-Point ICA is briefly introduced , implemented by Matlab and analyzed its feasibility in the blind source separation.The results from the experiment show Fixed-Point ICA algorithms will be useful to the ICA algorithm's improvement and its application in practice.Study how to make use of fixed ICA to effectively detect, separate and remove a wide variety of artifacts from EEG recordings, such as ECG EOG . Contrast experiments are respectively done by ICA and wavelet algorithms to remove the artifacts in the same EEG signals.The experimental results show the ICA can effectively remove the artifacts , almost do not lose the useful information in EEG signals.and is prior to the wavelet in the aspect keeping the detail of EEG signals.Fixed ICA is also applied to the analysis of Visual Evoked Potentials (VEP).The results show ICA can clearly separate VEP with physiological significance by few measurement time, and its effect is prior to the traditional superimposed averaging, which is useful to clinic.