Masking empirical mode decomposition-based hybrid features for recognition of motor imagery in EEG

Brain computer interface (BCI) in electroencephalogram (EEG) is of great important but it is challenging to manage the nonstationary EEG. In this paper, a newly-developed method, namely masking empirical mode decomposition (MEMD), is applied to deal with motor imagery (MI) recognition tasks. The windowed EEG is decomposed by MEMD and hybrid features are then extracted from the first two intrinsic mode functions (IMFs). Kruskal-Wallis test is carried out to extract significant features and they are finally fed into linear discriminant analysis (LDA) for classification. Results show our proposed approach can achieve the highest accuracy of 88.35% as well as the maximal mutual information of 0.6846. The presented technique is comparable or superior to other methods and is proven useful.