Design Of Virtual Rehabilitation System Based On Brain-machine Interface Of Motor Imagination

Motor Imagery Brain-Computer Interface(MI-BCI)directly converts the neural activity signals of the brain into control signals of a computer or external device,which can not only help people with limb movement disorders to effectively control external devices,but also provide a new strategy for rehabilitation treatment of stroke patients.The existing MI-BCI rehabilitation system has the defects of single scene mode,lack of feedback mechanism,poor individual adaptability and low recognition rate.Meanwhile,it is a challenge to maximize the deep activation and accelerate the functional remodeling mechanism of the motor nervous system.Thus,the purpose of this paper is to develop a multi-modal stimulation enhancement and neurofeedback strategy,which is applied to a virtual rehabilitation system based on MI-BCI,so as to meet the needs of limb rehabilitation training of stroke patients in flaccid paralysis period.The main contents of this paper are as follows:Firstly,the mechanism characteristics and acquisition methods of the Electroencephalogram(EEG)of the MI were analyzed,and EEG data is preprocessed and features were extracted in time,frequency and space domain.Considering the real-time requirement of motion intention recognition in rehabilitation system,the performance of support vector machine(SVM)and linear discriminator model were compared with the data of subjects,and the optimal intention resolution algorithm was determined.Secondly,a virtual rehabilitation platform based on multi-modal stimulation enhancement and multi-level neural feedback was designed and then built.Adopting the concept of layered design,this platform contains three core modules: EEG signals acquisition and intention recognition module,human-computer interaction and virtual scene management module,MySQL background database management module.It can realize multi-scenario rehabilitation training,rehabilitation assessment,data management,humancomputer interaction and other functions.In addition,aiming at the simple problem of feedback mechanism in the MI-BCI system,a multi-level visual neural feedback strategy was proposed.The characteristic frequency band of EEG was extracted by wavelet packet decomposition and the fourth order cumulative energy was calculated to evaluate the degree of motor nerve activation.Then,the spatial and temporal analysis of energy characteristics and the visual neural feedback were performed by using the brain topographic map.According to the individual adaptability needs of the subjects,a strategy of multi-mode virtual scene stimulation enhancement and parameter adaptive adjustment was proposed to ensure the best recovery effect of the current training scene.Finally,the effectiveness of the virtual rehabilitation system and multi-modal stimulation enhancement and neurofeedback strategy designed in this paper were verified.On the one hand,an experimental paradigm of virtual rehabilitation training and enhanced MI evaluation was designed to compare the changes in the energy characteristics of local frequency bands and intention recognition rate,and further to explore the neural activation mechanism of MI in different virtual scenes.On the other hand,the strategy and functional modules of the rehabilitation system were tested in subjects and applied to the hospital to complete the preliminary clinical verification.