Research On Electroencephalogram Classification Of The Brain-Computer Interface

A Brain-Computer Interface is a system for controlling or communicating with a computer orother electronic devices by human intentions. It’s mainly applied in the fields of medicine. Itprovides a new pattern for the patients who suffer from akinesia but have no injury with their headsto communicate with the outside world. In the research of BCI, the electroencephalogram (EEG)plays a very important role.And the key parts of the BCI system are feature extraction andclassification of the EEG signals.In the BCI, the feature extraction and classification of EEG can be achieved by massive studyof the Multilayer feedforward neural network. But the BP neural network based on error backpropagation converges slowly, and has low efficiency in training, limited accuracy in classification.To solve these problems, the quick and stable Levenberg-Marquardt algorithm is adopted in thisarticle instead of the BP algorithm to train the neural network. The MATLAB simulationexperiment about classifying the EEG signals of the motor imagery of left hand and right hand usesthe Graz data set B from the BCI competition2008. The simulation results show that the accuracyrate of this algorithm is87.1%, which is superior to78.2%of the BP algorithm, and it convergesbetter as well. This technology provides an effective way to EEG classification.BCI enables users to control devices with EEG activity from the scalp or with single-neuronactivity from within the brain. Both methods have disadvantages: EEG has limited resolution andrequires extensive training, while single-neuron recording entails significant clinical risks and haslimited stability. In the light of these problems, the electrocorticographic (ECoG) signals recordedfrom the surface of the brain can enable users to control a one-dimensional computer cursor rapidlyand accurately. The classification MATLAB experiment of the motor imagery of the left littlefingure and the tongue has reached a high classification accuracy of92%. This result reveals thatcompared to the EEG signals, ECoG signals can accurately locate the function cortex and avoid thechanges of amplitude, frequency and phase at the same time. In addition, our results suggest that anECoG-based BCI could provide for people with severe motor disabilities a non-muscularcommunication and control option that is more powerful and effective than EEG-based BCIs in thetwo-dimensional joystick movements.