The Study On The Automatic Removal Of Muscle Artifacts From Few-channel EEG

EEG is an electrophysiological monitoring technology,which has the characteristics of non-invasiveness,low cost and suitable for long-term monitoring.It has been widely used in the fields of medicine and science.Since the EEG signal is very weak,the collected EEG signal contains a large number of physiological artifacts,such as Electrooculogram(EOG),Electrocardiogram(ECG),and Electromyogram(EMG)artifacts.Due to the large amplitude and wide frequency range of muscle artifact,muscle artifact are the most difficult to remove among a variety of physiological artifacts.The existence of physiological artifact reduces the availability of EEG signals,so accurate removal of physiological artifact is essential for correct analysis of EEG signals.In recent years,wearable,real-time monitoring EEG recording equipment has become popular.The number of EEG channels collected by this type of equipment is relatively small.Common singlechannel and multi-channel artifact removal methods are not suitable for fewer channels.The characteristics of real-time monitoring of this type of equipment require that the artifact removal method should be automated and adaptive.Among the existing methods,the artifact removal method for the few channels has only been gradually proposed in recent years,and there are very few researches on the automatic removal of muscle artifact.In this topic,Least Square Multivariate Empirical Mode Decomposition and Interval Thresholding(LSMEMD-IT)is combined to realize automatic removal of muscle artifact in few-channel EEG.This method first uses LSMEMD to perform adaptive time-frequency decomposition of few-channel EEG signals to obtain Multivariate Intrinsic Mode Functions(MIMFs);then automatically calculates the muscle artifact threshold of the IMF based on the level of muscle artifacts in each IMF,use the improved interval thresholding function based on the Smoothly Clipped Absolute Deviation(SCAD)penalty and the muscle artifact threshold to remove muscle artifact in the corresponding IMF,and get artifact-free MIMFs;finally,use artifact-free MIMFs to reconstruct to get clean few-channel EEG.The experimental results of simulation and real data show that: Compared with the comparison method,LSMEMD-IT can realize automation,has better adaptability,retains EEG to the greatest extent and removes muscle artifact.Therefore,this subject believes that LSMEMD-IT is an effective and automatic method to remove EEG artifact in the case of few-channel.