EEMD-based selection of time-frequency patterns for motor imagery EEG
In this paper, a novel method based on Ensemble Empirical Mode Decomposition (EEMD) is presented to select time-frequency patterns for single-trial motor imagery EEG. The method comprises three progressive steps: 1) employ EEMD method to decompose EEG signal into a superposition of components or functions called IMFs, and then apply Hilbert transform to the IMFs to calculate the instantaneous frequency and instantaneous amplitude; 2) select the IMFs including the most useful information needed in classification; 3) select the time-frequency patterns according to the instantaneous frequency and amplitude of the selected IMFs. The proposed method was tested by using the EEG data of right and left motor imagery experiments from the BCI competition 2003. After selecting the time-frequency patterns, the features extracted by different methods are classified by Fisher linear discriminator. The results showed that the proposed method could improve the classification accuracy.
Author's Name: Gong, P., Chen, M., Zhang, L.
Volume: Volume 9
Issues: Issue 22
Keywords: Brain computer interface, Ensemble empirical mode decomposition, Motor imagery