Portable Motor Imagery Based Brain Computer Interface System Design

Motor imagery-based brain-computer interface(MI-BCI)technology has great value in motor control,neurological rehabilitation and intelligent control under the special environment.MI-BCI is expected to play an important role in improving the motion control ability,the information processing efficiency in brain and the rehabilitation of cerebral palsy due to the similar neural mechanisms with the actual movement.Many laboratories have developed online MI-BCI system which support real-time control command output,but due to the influence of several factors such as electroencephalogram(EEG)signal recording technique,not portable,operation is cumbersome,and etc,most of the developed online MI-BCI are only used in laboratory environments.For the past few years,FPGA-based,ARM-based,and other mobile platforms processors have a great improvement.Its algorithm processing and arithmetic speed have a great breakthrough.The mobile portable platforms have the ability to do most of the computing task of BCI systems.In order to make the MI-BCI more practical,we designed a portable MI-BCI system.The system use the multi-processor which is combine FPGA and ARM as core controller,and running the ARM Linux mobile platforms operation system.The system has two kind of MI task real-time control instructions and feedback.The system which is operation easily has a small size,and it can work with the rehabilitation training equipment such as exoskeleton robot,Electric wheelchair,and etc,to help patients improve theirs life quality and doing the rehabilitation training.The main work of this system includes portable system building,MI-BCI algorithm transplant and MI-BCI system program.The portable system building includes the work of the ADC controller,the IIR filter and the FIFO memory,all are based on FPGA,and reading data device driver program design.The algorithm transplant includes the~2,Common Spatial Patterns(CSP)feature extraction algorithm and Linear Discriminant Analysis(LDA)pattern classification algorithm,all are implement by C++.The MI-BCI system program includes a EEG signal display and a MI-BCI interactive program.It has the function of EEG data collection and display,task prompt and feedback,training task parameter estimation and online test,etc.Finally,for the sake of verify the portable system,we tested the EEG data collection,EEG signal display,IIR filter,MI-BCI algorithm and the whole system.The average accuracy of the MI-BCI system is 80.6%in the experiment of 15 subjects.The results show that the MI-BCI system has the same accuracy level of the system which is based on the PC platform,and satisfy the needs of applications.