Research On EEG Response And Performance Optimization Of Lower Limb MI-SSSEP Hybrid BCI

Motor imagery(MI)based on brain computer interface(BCI)can directly decode user’s motor intention and convert it into output instructions to control external equipment,which can provide new approaches for the rehabilitation of limb movement disorders.However,MI contains less information,which leads to lower performance of BCI system,and the combination of steady-state somatosensory evoked potential(SSSEP)and MI can effectively improve this limitation.Therefore,aiming at key problem of lower classification performance in lower-limb MI-BCI system,we verified that hybrid paradigm of MI-SSSEP can improve the performance of lower-limb BCI system firstly.And on this basis,different optimization methods were designed to further explore their impace on the performance of BCI system.Hence,we designed a hybrid paradigm(MI+ES),i.e.MI joint somatosensory electrical stimulation induced steady state somatosensory evoked potential(SSSEP),and compared with the traditional single paradigm(MI).EEG data of twenty-five healthy subjects were recorded.Fast fourier transform and event-related spectral perturbation(ERSP)was used to analysis the EEG feature response.The results showed that the cerebral central of subjects under two conditions showed event-related desynchronization in alpha band,and the somatosensory stimulation strategy under MI+ES conditions can also induce obvious SSSEP feature.Afterwards,the common space mode(CSP)/ filter bank CSP(FBCSP)was used to extract features,and the support vector machine(SVM)was used for pattern recognition to calculate the classification accuracy of the left and right foot MI under different conditions.The results showed that under the CSP algorithm,the average classification accuracy of the MI condition was 57.7%,and the MI+ES condition was 71.87%.The average classification accuracy of the FBCSP algorithm was 61.9% and 74.15%,respectively.It shows that the introduction of SSSEP could improve the performance of lower limb BCI system.In order to further explore the method of BCI performance optimization,experiments with different training modes were designed to research the effects of longterm off-line MI training on cortical activation response and BCI system performance.The EEG data of subjects with and without training and the same subjects before and after training were collected to explore the cortical activation response and the performance optimization method of lower limb MI-BCI system.The results showed that for the same subjects,MI training could significantly promote the activation of cerebral cortex and improve the MI-BCI performance of lower extremities under the MI condition.Which indicated the MI training method can improve the performance of single MI paradigm in lower limb BCI system.In summary,the superiority of MI-SSSEP hybrid paradigm in improving system performance was confirmed in this paper,which provides new technical ideas for the practical process of lower limb training MI-SSSEP-BCI system,and has a great application prospect.