EEG Separation Based On ICA

Electroencephalogram is the window for human to pry into the secret of the brain.So the thirst for finding out the mystery of our own drives the development of EEG in the whole century.Source localization in the human brain is known as an important problem of EEG. Source localization is useful for the therapy of neural pathologis such as epilepsy. In the cognitive science,it is the efficient method for researching all kinds of advanced neural activity and in the research of acupuncture mechanization it brings more and more important effect.But for the reason of the degree of problems fitting ill-posed being too much , the bottleneck of location befalls when locating multi-sources of EEG. Recently, a new technology of singal processing named Independent Component Analysis(ICA)is applied to the disposal of EEG.In this paper a new algorithm was designed which applied the technology of ICA to the analysis of EEG, and separated the voltages on the scalp from different sources using the independence of them then respectively estimating each source.Much simulation has been done in this paper, which proved the feasibility of the algorithm and provided bases for its actual application.The main studies of this paper included:(1) based on the theory of electric field,adopting the head model of three-concentric-shell,this paper established the forward-problem model of EEG(2) studied the theory and algorithm of ICA and compiled the MATLAB programs of FASTICA method of ICA, INFOMAX method and Principal Component Analysis method.(3) applied ICA to analysis of EEG and simulated, then testified the results can be applied to location.