抄録
In previous studies, approaches, which qualitatively determine the degrees and differences of chaos, like a calculation of fractal dimension and Lyapunov exponents, have been used as an electroencephalogram (EEG) analysis method. However, it is difficult to quantitatively grasp phenomenon appearing in EEGs and to demonstrate threshold by using this method.Therefore, proposing an EEG analysis method, which uses model parameters of nonlinear oscillator, we attempt to apply to quantitatively evaluate an EEG abnormality. Then, it is found that the proposed method could detect the difference between normality and abnormality. The aim of this paper is to clarify characteristics of parameters determined when the variation of ECoG would be simulated. The time series data, which change from normal ECoGs to epileptic ECoGs, were used to analyze. Extracting the time series data in alpha band, genetic algorithm was used as the proposed method need to solve the inverse problem. Each parameter is identified for analytical of 0.512 second and 1.024 second, and the variations for time are obtained. The values of the identified parameters which could related to nonlinearity, fall down in an epileptic state. Namely, at the epileptic EEG, the parameter corresponding to nonlinear term could be smaller than that at the normal EEG.
