A Method for Precise Extraction of EEG Rhythmic Activities Using Multivariate Empirical Mode Decomposition and the Hilbert Transform

Neural activations can be measured based on modulation of EEG rhythmic activities within specific frequency bands. Previous studies have typically extracted EEG rhythmic activities using Fourier based methods such as short time Fourier and wavelet transforms. But the methods tend to obscure intrawave frequency fluctuations and smear the energy over a much wider frequency range. In this study, we propose a method that combines multivariate empirical mode decomposition (MEMD) with the Hilbert transform, rather than a wavelet transform, to extract rhythmic activities more precisely and to visualize them more clearly. The performance of the method was validated using measured EEG data obtained by a wrist movement experiment. The results demonstrated that the proposed method can extract and visualize multi- channel EEG rhythmic activities with higher resolution than methods employing short time Fourier and wavelet transforms.