EEG signal analysis using spectral correlation function & GARCH model

作者：Mihandoost Sara^*; Amirani Mehdi Chehel

来源：Signal, Image and Video Processing, 2015, 9(6): 1461-1472.

DOI：10.1007/s11760-013-0600-9

• EEG signals
• SCF
• GARCH model
• LDA
• MLP classifier

In this paper, a new feature extraction method for electroencephalogram (EEG) signal analysis is suggested. This scheme is based on the spectral correlation SCF), which presents a second-order statistical description in the frequency domain. The SCF of each EEG signal is computed by using an efficient computational algorithm, which is called the FFT accumulation method. To choose an efficient statistical model for the SCF coefficients of EEG signals, their statistical properties are surveyed at different regions of bi-frequency plane. We show that the SCF coefficients are heteroscedastic, and the generalized autoregressive conditional heteroscedasticity (GARCH) is considered as one of their proper model. The GARCH parameters of each SCF sub-band are calculated and then are utilized for EEG classification. Hence, the resultant features fed to a multilayer perceptron classifier. The results clearly indicate that the performance of the new method in EEG classification outperforms the previous studies.

• 出版日期2015-9