Font Size: a A A

Research On Upper Limb Rehabilitation Based On Brain Computer Interface

Posted on:2015-09-17Degree:MasterType:Thesis
Country:ChinaCandidate:Y C MeiFull Text:PDF
GTID:2298330452453372Subject:Control engineering
Abstract/Summary:PDF Full Text Request
Stroke is one of the greatest threaten to the people s healthy for its high incidenceand disability rate. Approximately two-thirds of stroke patients are accompanied withupper limbs motor dysfunction, which brings great soul and physical pain to thepatients. So how to help them for post-stroke rehabilitation is becoming a hotspot inthe society.Brain-Computer Interface (BCI) is a communication control system which allowsdirect translation of brain states into actions, without the aids of usual muscularpathway. On account of neural plasticity, it is promising by applying the BCItechnology to the arm rehabilitation of stroke patients. And it can stimulate themovement tend, promote the motor dysfunction rehabilitation results. In this paper,studies were conducted from the aspects of ocular artifact removal, adaptive featureextraction, pattern classification and the on-line arm rehabilitation system. The mainachievements are listed as following:(1)The ocular artifact removal method based on DWT and CCAThe electroencephalography (EEG) is easily affected by the ocular artifact (OA),it brings difficulty to the following processing and application. A novel method wasproposed based on the combination of canonical correlation analysis (CCA) anddiscrete wavelet transform (DWT), and it is denoted as DWT-CCA. Firstly, DWT wasapplied to the collected EEG and electrooculogram (EOG) signals to acquire themultiple scale wavelet coefficients, and CCA to eliminate the correlation among thecoefficients. Then, the correlation coefficient was used as a criterion to recognize theocular components, and the corresponding canonical wavelet coefficient vectors wereset to zero. At last, the inverse algorithms of CCA and DWT were applied in sequence.So, the OA was removed from EEG in this way. By using DWT-CCA and othermethods, experiment research was finished based on the BCI data sets which contains4kinds of EOG data and sampled from9subjects at different time. The significanttests show that the proposed method has obvious superiority in the aspects of rootmean square error (RMSE) and signal noise rate (SNR). Furthermore, it has goodreal-time performance and excellent adaptive capabilities.(2)The recognition of EEG based on OHHT and ISVMThe motor imagery EEG/ECoG is very weak, and can be easily affected by the environments, experimenter condition, individual difference, and so on. Even more, itmay change during rehabilitation as time goes on. So it is a pivotal spot on how toextract and classify the adaptive features, which can reflect brain activities precisely.An adaptive EEG recognition method was proposed in this paper based on OHHT andISVM, denoted as OHISVM. Firstly, the ocular removal was conducted on EEGsignals, and the entropy was introduced to select the optimal combination of channels.Then, the OHHT method was applied to acquire the Hilbert instantaneous andmarginal energy spectrum of EEG signals, which were defined as the time andfrequency domain features. What more, the space domain feature was extracted basedon CSSD. A serial feature fusion technique was followed to obtain thetime-frequency-space domain features. At last, the ISVM was applied to therecognition of EEG signals. Experimental research was conducted based on the ECoGdate sets from International BCI Competition Database and the EEG signal sampledby our groups, associated with arm flexion/extension action. The result verified thepossibility and validity of this recognition method proposed in this paper.(3) The online arm rehabilitation system based on MI-BCIAn arm rehabilitation system was designed to control the arm robot, which actedas the recognition results of MI-BCI system. This rehabilitation system includes uppercomputer (PC) subsystem and lower subsystem based on the S3C2440A controller ofARM9. The upper computer subsystem works for EEG sampling and processing. TheEEG signals are sampled based on multi-threads technology and the C API functionsof g.MOBIlab. The processing parts consist of EEG artifact removal, featureextraction and pattern classification. The software on PC is developed based on MFC,and combined the mixed programming of MATLAB and C++. The lower subsystemworks for the control of arm robot and suggestion of arm movements on LCD. Andthe uc/GUI system was introduced to generate the suggestion frames. Communicationwas realized by serial port communication between the upper computer subsystemand the lower subsystem. It proved the feasibility of the arm rehabilitation systembased on MI-BCI. In this way, the initiative of stroke patients can be promoted, whichassociated with the increasing efficacy. The results show its potential applicationvalue and prospect in arm rehabilitation area.
Keywords/Search Tags:brain-computer interface, arm rehabilitation, ocular removal, orthogonalhilbert-huang transform, incremental support vector machine
PDF Full Text Request
Related items