Modeling of Electroencephalogram Parameters

Electroencephalography is widely used for diagnostic purposes, to detect diseases of the nervous system. Assessment of the patient’s condition is carried out according to the frequency composition of electroencephalogram (EEG). The classical Fourier transform method has insufficient resolution and does not allow to separate the frequencies that are close enough. At the same time, the separation of closely located frequencies is important in EEG analysis. In this paper, it is shown that parametric methods are good enough to separate frequency components of EEG. Using parametric models, more accurate estimates of power spectral density and higher frequency resolution are obtained than with classical spectral estimation methods. The autoregressive model and Prony’s method are considered. It is shown on the example of EEG realization that these models have better resolution compared to the classical Fourier transform.