Independent Component Analysis Based Signal and Noise Separation for Evoked Potentials
Qiu Tian-shuan · Beijing shengwu yixue gongcheng · 2003
The detection and analysis technology of evoked potentials(EPs) is an important means in clinical diagnosis for injury or disease of the central nervous system. Generally, the EP signals obtained from the body surface are always contaminated by heavy noises. The most typical one is the spontaneous electroencephalogram(EEG). Therefore, it is necessary to remove such noises like EEG from mixed signals in order to detect the injury or disease reliably. The independent component analysis(ICA) is a newly developed statistical signal processing method. In this paper we introduces how to remove noises in EP signals using an ICA based method, and the results are compared with the results obtained using traditional adaptive filtering. It is clear from computer simulations that the ICA based method is much more effective for signal and noise separation than the method of adaptive filtering.