Research On Bimodal Brain-computer Interface Based On Motor Imagery

At present,brain-computer interface(BCI)technology based on motor imagery has become a research hotspot in related fields.In order to study the brain activity mechanism of motor imagery and extract the characteristic quantity that can be used for the classification of BCI,the neural mechanism of motor imagery was explored by using the method of classification and recognition based on the study of EEG signals,the four-action motor imagery,and the dual modes of EEG and functional near-infrared spectroscopy.Firstly,the brain electrode channels that can be used for movement differentiation were screened in the dual-motion motor imagery experiment,and the two kinds of actions were classified.Secondly,four kinds of common actions were compared and analyzed in the four-action motor imagery experiment.The conclusion of movement differentiation was perfected in the time-domain and time-frequency domain.An command output architecture for recognition of four-action EEG signals was established based on support vector machine(SVM).Finally,by exploring the dual-mode motor imagery with the help of the existing experimental environment,the synchronous acquisition paradigm and method of EEG and NIRS were designed to evaluate the connection mode and differentiation degree of cerebral cortex in four states.In this paper,in a step-by-step way,the experimental paradigm was used to gradually deepen and improve the process.The neural mechanism of motor imagery was explored in the time domain,frequency domain and space domain,and the relevant characteristics of BCI were extracted for classification and evaluation.It was found that,based on the support vector machine classifier,the discrimination accuracy of two classifications under two actions could reach 75%,and the recognition accuracy of mixed four-action instructions could reach 62.47%.The energy values of the four actions were significantly different in the range of 0.2-10Hz.During the dual-mode imagery task,the brain activation region was basically consistent with the four actions,and the differentiation effect of NIRS signals was obvious between 10s and 12s.The information transmission of EEG and the change of blood oxygen concentration in NIRS could be explained by each other.