Collecting And Decoding The Cortical Neuronal Spike Signal Related To Stand-standing-Sit In Monkeys

Movement dysfunction caused by damaged nervous systems such as spinal cord injuries is a worldwide challenge to clinical treatments.Such disability causes severe life difficulties and psychological barriers for patients and also carries long-term burdens on families and society.Unfortunately,there has been no effective cure in clinical practice yet.It has been reported that decoding brain motor control mechanisms and developing intelligent neuro-prosthetic may provide new hopes for movement dysfunction patients to regain their athletic ability.The collection and analysis of the neuronal spike signal of the primary motor cortex(M1)is the key to decode the brain motor control mechanism.It is also the basic requirement for the development of a new generation of intelligent neuro-prosthesis.Currently,intelligent neuro-prosthesis mainly focus on upper limb research,and there are many related achievements.Especially,the research on decoding the impulse signals of cortical neurons related to the upper limb movement by using neural network technology based on the recursive Bayesian algorithm has been a great success.However,the study of the brain controlling the lower limb movement faces technical challenges.One of the reasons is that the brain area related to the lower limb movement is surgical untouchable,for it is in front of the intersection on the central sulcus and the top suture,an area with rich blood vessels.Another reason is the lower limb movement on animal behaviors still lacking proper experimental design.Therefore,the application of cortical neuronal spikes to decode lower limb movement has not been reported yet.This topic brings a new insight to develop an intelligent neuro-prosthetic.In this work,the method of recording electrodes was specially designed and the sinusoidal region of the lower limbs was avoided by oblique insertion.Hence,the cortical signals related to the lower limb movement were successfully collected in the monkey brain.Additionally,this work also developed a special stand-standing-sit task for monkeys.In a virtual environment,monkeys were trained to perform the stand-standing-sit task according to the command operation.Electromyography(EMG)of main muscles,the trajectories of the lower limbs(Kinematics)and cortical neuronal spikes from the M1 in the monkeys were recorded.The Kalman filter(KF)algorithm and unscented Kalman filter(UKF)algorithm were applied to decode EMG of the main muscles and kinematics of the lower limb.The muscles included soleus,tibialis anterior,semitendinosus,rectus femoris,long extensor digitorum,long flexor digitorum,and the long flexor.The amplitude of the EMG signal was considered to be the system state,and the cortical neuronal spikes were regarded as the observation for the state of the system,and the corresponding frequency was immediately issued according to the action potential observed in real-time.The amplitude of the myoelectric signal envelope for each muscle in the lower limb was predicted.Furthermore,this work also used the KF and UKF algorithm to predict the trajectory,velocity,and acceleration of the monkey's right leg ankle joint.There were total 1598 cortical neurons recorded in 2 monkeys' M1 regions,in which 294(18.4%)neurons encoding initiation of standing,310(19.4%)neurons encoding standing up,104(6.5%)neurons encoding standing hold,and 205(12.8%)neurons encoding sitting down.The correlation coefficient(CC)showed that the UKF algorithm had a higher percentage(9.05%)than the KF algorithm in predicting the EMG signal of the main lower limb muscles.The signals to noise ratio(SNR)had the same results,the UKF algorithm had a higher percentage(17.37%)than the KF algorithm in predicting the EMG signal of the main lower limb muscles.In addition,the UKF algorithm also showed the higher CC value and SNR value in reconstructing the trajectory of the lower limb movement with the cortical spike signal,though the velocity and acceleration showed lower precision.All these results suggested that both the KF algorithm and the UKF algorithm are suitable to decode lower limb EMG and predict the trajectory with cortical spike signal,and the UKF algorithm is the better alternative.