Ai Lingmei1,Huang Liyu1,Huang Yuangui2,Wang Jue1
(1.Institute of Biomedical Engineering,Xi'an Jiaotong University,Xi'an 710049,China;2.Department of Neuropathology,Xijing Hospital,Fourth Military Medical University,Xi'an 710032,China)
Abstract:Based on the fact that the signals of epileptic electroencephalogram (EEG) possess non-Gaussian and nonlinear stochastic properties,the higher-order statistic methods were used to estimate the bispectrum of epileptic EEG.The Guassian deviations of the EEGs in different stages of seizure were studied, in order to obtain more sensitive and accurate parameters for clinical epileptic monitoring and seizure prediction.By using the parameter model method to estimate bispectrum of EEG,model parameters are determined based on the singular value decomposition least square method. Thereby the bispectrum estimation with a high resolution and some available EEG phase information can be obtained. The higher order spectrum is normalized with conventional power spectrum, and the result show can be used as coherence coefficient, and its relation with epileptic seizure can be found by estimating the coefficient.The experimental results show that contour diagrams of bispectrum possess obvious peaks in paroxysm,so and beforebegin paroxysm of epileptic,and bicoherence coefficient of EEG is very high beforeafter paroxysm,so it boosts up nonGaussian and nonlinear of brain wave.It can offer some new ideas for research of epileptic EEG and they may be potential parameters for clinical epileptic monitoring.
Keywords:epilepsy;electroencephalogram;bispectrum