Extraction Of Auditory And Visual Evoked Potential Based On Rough Sets Theory And Informax ICA Algorithm Research

Evoked potentials are the central nervous system response to external stimuli to make electricity. The signal of the auditory evoked potential contains information has an important significance for disease diagnosis, evaluation of human neurological diseases of the nervous system etc.. With weak amplitude, strong nonlinear, strong coupling and low SNR signal features, which make extract the 0.1uV evoked potentials in strong interference environment become a frontier research topic at domestic and abroad.At present, clinic widely used is the average superposition technique, this technique is simply and convenient. However, it need to test many times and lead to fatigue easily; and the loss of a large amount of dynamic information cannot accurately describe the dynamic characteristics of auditory evoked potential. As the field of physiological signals years of continuous development, a variety of different algorithms are used to extract the evoked potentials in-depth study. Wavelet threshold denoising method, set up the artificial neural network nonlinear approximation model of evoked potentials, high signal to noise ratio of independent component analysis algorithm, wavelet transform and independent component analysis in combination with the WICA algorithm, wavelet neural network algorithm, etc.This article focuses on the use of rough sets theory combined with Informax ICA method to extract the brainstem auditory evoked potential and reverse pattern visual evoked potential. Acquisition of evoked potential signal to noise ratio is low, contains a lot of frequency signals are higher than the evoked potential and mixed with evoked potential is the same frequency band and overlapping strong noise signal. For Informax ICA cannot effectively extract the low SNR signal, this paper first acquire the signal preprocessing in rough set theory, improve SNR of evoked potential, after pretreatment signal Informax ICA algorithm isolated bands overlapping evoked potentials and noise and extract evoked potential waveform obtained. When the signal preprocess, according to the features of auditory and visual evoked potential waveform, the time interval to pole threshold is set to 0.45 ms and 0.69 ms respectively to remove part of the high frequency noise. Through the analysis of rough set preprocessed signal, the signal spectrum, power spectral density estimation, the SNR and the mean square error, and compared with the average superposition signal preprocessing, rough set theory is more efficient than the average removal of brainstem auditory evoked potentials and visual evoked potential reversal pattern in the high-frequency noise, improve SNR. Finally combine waveforms of evoked potential respectively with average superposition technique and Informax ICA and algorithm based on rough sets theory and Informax ICA. Through analysis, compare the three algorithms of brainstem auditory evoked waveforms â… -â…¢, â…¢-â…£/â…¤ and â… -â…£/â…¤ peak to peak latency period and the time of P100 of reverse pattern visual evoked potential, based on rough set theory and the method of combining Informax ICA can more truly and effectively extract the brainstem auditory evoked waveform and reversal pattern visual evoked potential waveform, and provide a technical mean for clinical medicine, a new idea for extracting somatosensory and the visual evoked potential.