Early Decoding of Tongue-Hand Movement from EEG Recordings Using Dynamic Functional Connectivity Graphs

In this paper, we consider the problem of early decoding of EEG signals associated with tongue-hand motor execution and imagery tasks. A two-step feature extraction strategy is proposed. In the first step, the EEG data is segmented into sequences of quasi-stationary intervals, during which functional networks sustain their connectivity. The second step localizes the functional networks during each segment, to generate the dynamic functional connectivity graphs. Next, the common spatial pattern (CSP) algorithm is employed to select features for the classification problem. To take advantage of the dynamic nature of the generated graphs, a long short term memory (LSTM) classifier is used for classification. Using the first 500 ms of EEG recordings, the proposed framework is capable of classifying different tongue-hand motor execution and imagery tasks, with an average accuracy of 79%.