Real Time Brain-Computer Interface Design And Study

Brain-Computer Interface (BCI) formed in1970s has become more and morepopular and has got fast development. It doesn’t rely on the normal output channel ofour brain such as the Peripheral nervous system and muscle tissue, but on thecommunication system which combines our brain with computer or other electronicequipments. BCI provides users, especially those patients who have dyskinesia butnormal brain function, a good approach to the external world and thus increases theirlife quality. What’s more, it has the potential application value in the fields of industry,aviation and military. This paper targets at the design of real time BCI system especiallyaims at achieving the exigent communication rate, processing rate and the accuracy rate.In this paper, we use the LabVIEW, a program language, to set up a real timetransmission and processing system. After studying the type of TCP connection datacell, the encapsulation protocol, and the connection procedural, we have designed theinterface circuit and the communication protocol, which supports the realization of thereal time cursor control system and the control of the remote control car. The resultsshow that both systems have achieved real time control and even ensure high accuracyrates.The signal feature extraction and pattern recognition are very important in BCIsystem. This paper is about the study of BCI system based on the right and left handmotor imagery. First, we have done the off-line analysis by using the right and left handmotor imagery data of BCI2003. After analyzing the power spectrum of signals fromC3and C4channel, we recognize that μ rhythm is better for the feature extraction. Thenwe use AR model to analyze individual’s electroencephalogram(EEG) signal of motorimagery, and by comparing the frequency spectrograms in different order of the model,we find that12th order AR model is much better for us to do the feature extraction thanothers. At last, we use the Fisher Discriminatory Analysis (FDA) as our classifier to dothe pattern recognition, and finally reach a high recognition accuracy rate, which givesthe theoretical and experimental supports for the study of motor imagery basedasynchronous BCI.First, the off-line analysis of EEG has been to determine method of real-time BCIfeature extraction in this paper. This paper achieves building a real time system basedon α rhythm and eye electrical signal, which provides our research group an experimental platform for the study of BCI. This paper also realizes real time motorbased synchronous BCI, which provides much experience for the further study of realtime BCI system.