Research On Arm Rehabilitation Based On Motor Imagery Eeg

According to statistics, the Chinese aging population is more than180millionand increasing with8million per year, the aging problem is becoming increasinglyserious. In addition, stroke has become one of the greatest threats to old people’shealth, and most patients suffer from hemiplegic paralysis. This brings a heavy burdento the social and their family. So how to help the patients for post-stroke rehabilitationis urgent as well as difficult in our society.Brain-computer interface (BCI) is a communication control system which allowsdirect translation of brain states into actions, bypassing the usual muscular pathways.Motor imagery (MI) mental tasks can revive the dormant synapses to compensate forbrain damage after stroke, and therefore restore the impaired cortical areas. So theBCI technology with MI (MI-BCI) can promote cortical plasticity and provide anactive way for post-stroke rehabilitation. In this paper, the analysis and processingmethods for motor imagery electroencephalography (EEG) and electrocorticography(ECoG) were finished, and a MI-BCI based arm movement function rehabilitationsystem was designed for stroke patients. The main results are as follows:(1) The ocular artifact removal method based on DWT and ICAOcular artifact (OA) is one of the main interferences in EEG recording andbrings about much difficulty in further signal processing. Based on discrete wavelettransform (DWT) and independent component analysis (ICA), a novel automaticmethod of OAremoval, denoted as DWICA, was proposed. The angle cosine criterionwas introduced to recognize OA component, then experimental data contaminatedwith OA was constructed according to bi-directionality between EEG and EOG,several quantitative performance indexes were computed to assess DWICA.Experiment results show that DWICA is preferable and effective in automatic OAcorrection. Meanwhile, it is powerful in noise immunity and fast in convergence rate.This provides the foundation for EEG preprocessing online in BCI rehabilitationsystem.(2) An adaptive feature extraction method with multi-domain fusion strategybased on HHT and CSSDThe motor imagery EEG/ECoG may change during rehabilitation as time goes on,an adaptive feature extraction method, denoted as HCSSD, was presented. Firstly, arelative distance criterion was defined to select the optimal combination of channels. Secondly, Hilbert instantaneous energy spectrum and marginal energy spectrum ofEEG/ECoG were calculated to extract time feature and frequency feature respectively.Then CSSD was applied to extract spatial feature. Furthermore, serial feature fusionstrategy was adopted to obtain time-frequency-spatial feature. For the motor imageryECoG data, the average recognition accuracy was92%with one week interval inrecording the training set and test set. Experiment results show that HCSSD canenhance the adaptivity of feature extraction, with the recognition accuracy improved.So this method provides a new idea for BCI application in rehabilitation field.(3) The experiment design and analysis of motor imagery EEGTwo experiment schemes were designed for motor imagery EEG recording andoffline analysis, including left little finger/tongue and arm extension/flexion motorimagery movement. The corresponding collecting software was programmed withWin32forms application, g.MOBIlab+API and OpenGL, providing a goodhuman-computer interaction interface for EEG recording. Then DWICA and HCSSDwere applied to process EEG data to validate the theory methods in practice. Finally,the characteristics of arm extension/flexion motor imagery EEG signals were studiedbased on neurophysiology. It provides a fundamental basis for the design of onlineMI-BCI based arm rehabilitation system.(4) The design and realization of MI-BCI based arm movement functionrehabilitation systemAn online MI-BCI system was developed for arm movement functionrehabilitation and its key parts were researched in detail. The real-time recording andprocessing software was achieved by hybrid programming and multithreadingtechnology. And then the control module was designed for mechanical arm based onS3C2440A. Finally, the request response communication protocol between controlmodule and PC was defined to control the corresponding extension/flexion action ofmechanical arm during motor imagery. This preliminary rehabilitation system isexpected to realize the active therapy for arm paralysed patients, and experimentresults prove its feasibility and potential application value.The research is helpful to improve the reliability and adaptivity of BCI system,and may promote the application of MI-BCI in rehabilitation field.