A Research On The Methods Of Preprocessing For EEG Signals Source Location

Scalp EEG has been used as an important non-invasive approach for research in the quantitative study of human cerebral physiological and pathological functions, and used as a strong tool for the comprehensive applications in the preclinical medicine, clinical medicine and medical engineering. It is critical to study the preprocessing methods on EEG signals for the preclinical medicine research and medical engineering applications.This dissertation first introduces the physiological and physical foundation, classification and acquirement of EEG signals besides the international 10-20 system, then the corresponding preprocessing strategies are given.Secondly the basic theories and algorithms of independent component analysis (ICA) are studied in detail. The performance and characteristic of ICA algorithms are analyzed in succession.Based on the fact that multichannel EEG signals satisfy the fundamental provided conditions for ICA,a modified fast ICA algorithm is given and proved to be reliable by means of simulation. Furthermore, the algorithm is utilized to preprocess the real EEG signals. The experiment results confirm the validity of the algorithm.In the end, we discuss wavelet transform(WT) and its application in EEG signal preprocessing. Combining EEG signal nature with the multiresolution analysis of wavelet transform, different wavelet denoising methods are used to preprocess EEG signal with different wavelet bases, and the choice of best wavelet basis is discussed. Then the effectiveness of different wavelet methods is compared with each other, providing the technological references for EEG signal preprocessing with wavelet transform.