New blind estimation method of evoked potentials based on minimum dispersion criterion and fractional lower order statistics

Daifeng Zha

doi:10.4236/jbise.2008.12015

Abstract

Evoked potentials (EPs) have been widely used to quantify neurological system properties. Tra-ditional EP analysis methods are developed under the condition that the background noises in EP are Gaussian distributed. Alpha stable distribution, a generalization of Gaussian, is better for modeling impulsive noises than Gaussian distribution in biomedical signal proc-essing. Conventional blind separation and es-timation method of evoked potentials is based on second order statistics or high order Statis-tics. Conventional blind separation and estima-tion method of evoked potentials is based on second order statistics (SOS). In this paper, we propose a new algorithm based on minimum dispersion criterion and fractional lower order statistics. The simulation experiments show that the proposed new algorithm is more robust than the conventional algorithm.

Highlights

The brain evoked potentials (EPs) are electrical responses of the central nervous system to sensory stimuli applied in a controlled manner
Its main applications re in feature extraction, blind source separation, biomedical signal processing
We present the basic data model used in both PCA and the source separation problem plotted in Figure 2, and discuss the necessary assumptions

Summary

Introduction

The brain evoked potentials (EPs) are electrical responses of the central nervous system to sensory stimuli applied in a controlled manner. Signal processing techniques including adaptive filtering, three-order correlation, and singular value decomposition (SVD) have been used in fast estimation of EPs. Independent component analysis (ICA) appeared as a promising technique in signal processing. Recent studies [5, 6] show that alpha stable distributions is better for modeling impulsive noise, including underwater acoustic, low-frequency atmospheric, and impulsive EEG,ECG, than Gaussian distribution in signal processing. Often the EEG signals are assumed to be Gaussian distributed white noise for mathematical convenience. EP analysis algorithms developed under the Gaussian EEG assumption may fail or may not perform optimally. An analysis shows that the alpha stable model fits the noises found in the impact acceleration experiment under study better than the Gaussian model [8]

Objectives

Results

Conclusion