Enhanced Motor Imagery-Based Eeg Classification Using A Discriminative Graph Fourier Subspace

Dealing with irregular domains, graph signal processing (GSP) has attracted much attention especially in brain imaging analysis. Motor imagery tasks are extensively utilized in brain-computer interface (BCI) systems that perform classification using features extracted from Electroencephalogram signals. In this paper, a GSP-based approach is presented for two-class motor imagery tasks classification. The proposed method exploits simultaneous diagonalization of two matrices that quantify the covariance structure of graph spectral representation of data from each class, providing a discriminative subspace where distinctive features are extracted from the data. The performance of the proposed method was evaluated on Dataset IVa from BCI Competition III. Experimental results show that the proposed method outperforms two state-of-the-art alternative methods.

1. INTRODUCTION

Electroencephalogram (EEG) is a non-invasive, high temporal resolution brain imaging modality that captures functional and physiological changes within the brain [1], [2]. Motor imagery (MI) tasks are dynamic states during which neuronal activity in the primary sensorimotor areas modifies similar to a real executed movement [3]. MI tasks are extensively utilized in brain-computer interface (BCI) systems and can be classified by extracting features from EEG signals to identify a user’s mental state [4], [5]. Many methods have been proposed to classify MI tasks from EEG signals. Some approaches are based on extracting key information from the time and frequency domains [6]-[8]. Some other approaches of MI classification are focused on learning spatial filters from multichannel EEG signals to extract discriminative features from data [9], [10]. There are also many studies which have proposed applying mathematical transforms, such as wavelet transforms, to extract discriminative features via decomposition of EEG signals [11], [12].