Research On Decoding Single Hand Continuous Movement Parameters And Arm Muscle Activities Based On EEG Signals

It is an emerging frontier of research on the use of neural signals for prosthesis control,in order to restore lost function to amputees and patients after spinal cord injury.Compared to the invasive neural signal based brain-machine interface(BMI),a non-invasive alternative,i.e.,the electroencephalogram(EEG)-based BMI would be more widely accepted by the patient populations above.Ideally,a real-time continuous neuroprosthestic control is required for practical applications.However,conventional EEG-based BMIs mainly deal with the discrete brain activity classification.Until very recently,the literature has reported several attempts for achieving the real-time continuous control by reconstructing of the continuous movement parameters(e.g.,speed,position,etc.)from the EEG recordings,and the low-delta band EEG is consistently reported to encode the continuous motor control information.Previous studies have extensively relied on the amplitude representation of such slow oscillations of EEG signals for building models to decode kinematic parameters.Inspired by the recent successes of the utilization of instantaneous phase of low-frequency invasive brain signals in the motor control decoding and sensory processing domains,this study examines the extension of such a slow-oscillation phase representation to the reconstruction of two-dimensional hand movements,with the non-invasive EEG signals for the first time.The data for analysis are collected on five healthy subjects performing 2D hand center-out reaching along four directions in two sessions.On representative channels that are known to show the execution information of reaching movements,the analysis show that the low-delta EEG phase representation is characterized by higher signal-to-noise ratio and stronger modulation by the movement tasks,compared to the low-delta EEG amplitude representation.Furthermore,we have tested the low-delta EEG phase representation with two commonly used linear decoding models(i.e.,multiple linear regression and Kalman filter).The results demonstrate that the low-delta EEG phase based decoders lead to superior performance for 2D executed movement reconstruction to its amplitude based ones,as well as the other-frequency band amplitude and power based ones.Thus,our study contributes to improve the movement reconstruction from EEG by introducing a new feature set based on the low-delta EEG phase patterns,and demonstrates its potential for continuous fine motion control of neuroprostheses.EMG contains abundant limb movement information,flexion and extension force and joint torque and so on.EMG has been widely used as a physiological feedback signal for the control of external power supply devices.Therefore,decoding muscle activity from EEG signals(i.e.time-domain features of filtered EMG signals)not only provides the possibility to help patients with high amputation or spinal cord injury to achieve motion control of external auxiliary equipment,but also provides the possibility to provide force control,which is conducive to the development of a new type of BCI similar to healthy people who can exercise and force dual control of external equipment.Firstly,this paper designs a discrete gesture experiment and a grip force experiment,and simultaneously records the electromyographic signals and electroencephalographic signals generated in the two experiments.Then preprocess EEG and obtain the time-domain features of EMG,explore the four kinds of decoding models(multiple linear regression and BP neural network,echo state network,XGBoost)in the performance of the decoder.The results show that under two kinds of experimental condition,XGBoost has the best excellent decoding performance.In addition,the influence of lead time of EEG and EEG electrodes on decoding results is investigated.Finally,this paper compares the classification performance of original EEG,EMG and decoded EMG signal in gesture recognition and grip strength experiment,and concludes that decoded EMG,compared with the original EEG,can help to improve the recognition and classification rate of gesture and grip strength,which shows the potential of the proposed method in the application of nerve prosthesis strength control.