Speed Imagery EEG Classification with Spatial-temporal Feature Attention Deep Neural Networks

Decoding continuous brain intentions is a major challenge for the research and application of brain-computer interfaces (BCI). Neuronal activity has been experimentally observed through various brain activity measuring techniques, of which electroencephalography (EEG) is the most widely used as it is noninvasive, practical, and has high time resolution. Here we propose a spontaneous speed imagery BCI paradigm with an EEG signals decoding method, in which a spatial-temporal feature attention deep neural network is developed to decode the continuous brain intentions. The speed imagery EEG signals of 0 Hz, 0.5 Hz and 1 Hz of left-hand clenching by 11 healthy subjects are decoded in experiments. The results reveal that the proposed method has the advantages of good performance and high efficiency, which is of great significance for patient rehabilitation and consumer applications.