Application of the Empirical Mode Decomposition to the Extraction of Features from EEG Signals for Mental Task Classification

In this work, it is proposed a technique for the feature extraction of electroencephalographic (EEG) signals for classification of mental tasks which is an important part in the development of Brain Computer Interfaces (BCI). The Empirical Mode Decomposition (EMD) is a method capable to process nonstationary and nonlinear signals as the EEG. This technique was applied in EEG signals of 7 subjects performing 5 mental tasks. For each mode obtained from the EMD and each EEG channel were computed six features: Root Mean Square (RMS), Variance, Shannon Entropy, Lempel-Ziv Complexity Value, and Central and Maximum Frequencies, obtaining a feature vector of 180 components. The Wilks' lambda parameter was applied for the selection of the most important variables reducing the dimensionality of the feature vector. The classification of mental tasks was performed using Linear Discriminate Analysis (LD) and Neural Networks (NN). With this method, the average classification over all subjects in database was 91 +/- 5% and 87 +/- 5% using LD and NN, respectively. It was concluded that the EMD allows getting better performances in the classification of mental tasks than the obtained with other traditional methods, like spectral analysis.