Online Recognition Of MI-SSSEP In Virtual Reality Environment

The number of people losing limb Motor function caused by chronic stroke and other diseases is increasing year by year.The possibility of spontaneous recovery of these diseases is very small.The Brain Computer Interface based on Motor Imagery(MI),BCI)system can improve movement function in patients with hemiplegia after stroke,help patients for motor function recovery,but exercise now imagine brain-machine interface design paradigm is relatively simple,requires a lot of training to improve the control ability of the user,and the effect is not obvious in the process of training,and then the system recognition rate is not high.In order to improve the classification and recognition rate of motor imagination EEG signals,the steady-state somatosensory evoked potential(SSSEP)features were fused with motor imagination on the basis of motor imagination.To study the characteristic distribution and classification recognition accuracy of lower limb EEG signals in time-frequency domain and spatial domain assisted by unilateral(left and right foot)and bilateral electrical stimulation in mi-SSSEP composite paradigm.Unilateral left foot electrical stimulation mode(only left ankle posterior tibial stimulation assisted),unilateral right foot electrical stimulation mode(only right ankle posterior tibial nerve assisted)and bilateral simultaneous electrical stimulation mode(both left and right ankle posterior tibial stimulation assisted)were designed.The results showed that the average classification accuracy of simultaneous foot stimulation was 70.13%,and that of unilateral right foot stimulation was 75.70%.According to the T-test,the difference in classification accuracy of simultaneous electrical stimulation on both feet versus unilateral electrical stimulation on the right foot was statistically significant(P=0.03143< 0.05),proving that reducing electrical stimulation assistance on one limb could improve the classification and recognition rate.The effect of Virtual Reality(VR)environment guidance on the performance of MI-SSSEP eeg signal recognition was studied.The combination of VR technology and MI can provide "immersive" virtual training scenes,which can greatly improve the training enthusiasm of patients and improve the efficiency of rehabilitation while reducing the cost of rehabilitation training.In the experiment,MI and SSSEP features were fused and VR technology was combined to design two experimental modes: virtual reality environment(HVR)and virtual reality environment(NVR).The results showed that the average classification accuracy of 10 subjects in HVR stimulus mode was 81.38%,and that of NVR stimulus mode was 75.75%.According to the paired T-test,the difference of classification accuracy of the two stimulus modes was statistically significant(P=0.035543< 0.05),and the study proved that the classification accuracy of the composite MI-SSSEP paradigm was higher in VR environment.Based on the offline system,the study designed an online eeg acquisition system combining VR environment guidance and MI-SSSEP composite paradigm to help stroke patients with rehabilitation training.The experimental design is divided into offline training stage and online feedback stage.The online acquisition system based on VR environment can collect the eeg signals of the subjects in real time,and show the prompt actions in the form of animation in the virtual reality environment.The subjects can compare and adjust the actions according to the feedback of virtual characters in real time.The results show that the average classification accuracy of the online collection system is 80.75%,and the variance of the overall average classification accuracy is 6.876894.The average ITR of 10 subjects is 17.12bits/min,which proves the feasibility of the online collection system in rehabilitation.