An Embedded Brain Machine Interface System Based On ECoG Signals

There are many patients in the world who have suffered inconvenience due to the lack of motor function,it brings them various kinds of inconvenience and seriously impacts on their living conditions and quality.Unfortunately the drug or the surgical treatment effect is not obvious.The Brain Machine Interface provides a new means of treatment and rehabilitation for these patients.But the traditional Brain Machine Interface system is composed of commercial collector and PC,which is bulky and expensive.It is not conducive to its popularization and application in clinical and practical life.Therefore,based on the characteristics of embedded system,and the research of the original Brain Machine Interface technology,this paper designs and constructs a complete embedded Brain Machine Interface system,and completes the online asynchronous control of ECoG signals in patients to external manipulators on this system.The proposed embedded Brain Machine Interface system mainly includes two parts:the front end signal acquisition chip RHD2132 and the back-end embedded ARM processing platform Jeston TK1.First of all,the ARM processing platform as the core of the system,through its high-speed SPI interface connected to the front of the acquisition chip,control signal acquisition start and stop,and signal acquisition related parameters configuration,including the sampling rate,bandpass filter cut-off Frequency,sampling channel selection and so on.Then,the processing platform receives the original signal sent by the acquisition chip,to complete all the functions including the data preprocessing,feature extraction,signal decoding and robot control in its CPU.Data preprocessing includes the spatial filtering which gets rid of the noise and smooths the data.Feature extraction includes time and frequency conversion and choose the specific spectrum in frequency as the input of the classifier.Signal decoding is based on the SVM classifier,we design an asynchronous decoding algorithm which is based on two stage decoding;The final decoding output is converted to the corresponding control instructions and is passed to the external manipulator to complete control of the manipulator.Experimental results showed that the proposed embedded Brain Machine Interface system can accomplish all the expected functions,and there was no obvious difference of the decoding accuracy compared with the traditional Brain Machine Interface system.But it had the obvious advantages on the volume of system,portable degree and the system power consumption,which could well meet the needs of clinical rehabilitation and real life applications.In conclusion,the proposed embedded Brain Machine Interface system has high reference value for research and lay a good foundation for subsequent development.Also,it proves the value and significance for the future development of embedded Brain Machine Interface system.