Research On Brain-Computer Interface Paradigm Of Somatosensory Channels

Brain Computer Interfaces(BCIs)provide a new communication and control channel that does not depend on brain's output pathways of peripheral nerves and muscles.Evoked potential-based BCI(EP-BCI),which associates with advanced thinking activities to be expressed and external stimuli,is a branch of BCI research whose evoked potential is more concentrated in a certain period or a certain frequency range.According to the different stimulus frequencies,evoked potential can be divided into event related potential and steady-state evoked potential.Usually,EP-BCI depends on visual channel.However,the applicability of EP-BCI based on this channel is limited for patients with impaired visual function or whose visual channel is occupied by other tasks.And somatosensory,due to wide distribution and strong sensitivity,has become an important channel to replace visual EP-BCI.Accordingly,based on tactile Event Related Potential(tERP)and Steady-State Somatosensory Evoked Potential(SSSEP),we explored and optimized these two paradigms.Firstly,the tactile stimulus module with adjustable intensity and frequency was designed independently.Under the condition of less than 4 pain channels,the stimuli were output steadily through the platform,based on which,the feasibility of tERP and SSSEP were validated.To ensure stable operation under multiple output channels,we have designed a multimodal programmable tactile stimulus system that meets the requirements of this experiment on the basis of existing hardware.The system has strong programmability,good extensibility and simple operation.Secondly,we propesed a novel paradigm—sequential omitteded paradigm.In the traditional tERP sequential paradigm,each motor vibrates one by one following random order with a interval between the adjacent two vibrations.The classification accuracy of tERP is equal to that of visual P300.However,the information transmission rate is less than half.The sequential omitted paradigm,based on the inherent characteristics of tactile stimulation,increases the rate of information transmission by reducing or even canceling the interval of stimulation under the premise of ensuring that the accuracy of classification is not reduced.This paradigm has doubled the rate of information transmission of the current tERP-BCI system,exceeding 20 bits/min,and ensuring that the online classification accuracy is around 85%.Thirdly,normalized-Power Spectral Density Analysis(N-PSDA)was innovatively come up with.Compared with the traditional PSDA,this method decreases the adverse effects brought by low signal-to-noise ratio and large variance of different stimulation magnitude efficiently and improves the classification accuracy of each stimulus frequency significantly.Lastly,the author designed experiment--by means of tERP paradigm,to control mechanical arm with six degrees of freedom in realization of object capture task.The performance of five subjects revealed that the somatosensory channel EP-BCI has a stable performance and strong interactivity.The applicability of the control system in the actual multi classification task is enhanced via this kind of BCI.