The Study On Muscle Artifact Removal Methods Towards Wearable EEG Devices

As a representative technique of reflecting electrical activities generated by the cerebral cortex,electroencephalography(EEG)is widely adopted for numerous practical applications in the biomedical engineering field.It owns the benefits of low cost,easy usability,high temporal resolution and so on.However,EEG signals collected from human scalps are often polluted by diverse artifacts,for instance electromyogram(EMG),electrooculogram(EOG),and electrocardiogram(ECG)artifacts.Among them,EMG artifacts are featured with high amplitude,non-stereotyped scalp topographies,as well as widespread spectrum distributions,causing the noise reduction most challenging.Recently,wearable EEG recording devices are one major trend of EEG acquisition equipment.In the mobile situation,EEG data are inevitably contaminated by muscle activities and meanwhile a limited number of channels are preferred,making muscle artifact removal more difficult.At present,several researchers have proved the superiority of combining singlechannel decomposition algorithms with blind source separation(BSS)to make EEG recordings free from EMG contamination.In this study,we come up with a valid method to accomplish muscle artifact removal from EEG by using the combination of singular spectrum analysis(SSA)and canonical correlation analysis(CCA),which is named as SSA-CCA.SSA-CCA utilizes SSA to decompose each channel into a collection of reconstructed components(RCs)and then applies CCA to perform the subsequent noise reduction process.Based on superior decomposition of SSA and the two-step strategy of combining two kinds of methods,the proposed method can flexible use different algorithmic frameworks and achieve the best effect on muscle artifact removal from EEG so far for the three situations(i.e.multichannel,few-channel and single-channel EEG).The performance of SSA-CCA is evaluated on semi-simulated data and real epilepsy data.The results demonstrates that this work is a promising approach owning both the effectiveness and low time cost,thus suited to real-world wearable EEG devices with one or a few channels.Moreover,in order to facilitate the researchers without engineering background removing EMG artifacts from EEG,this study develops an open-source toolbox that integrates most state-of-the-art methods through a user-friendly Graphical User Interface(GUI),including the proposed SSA-CCA method of this article.There are two kinds of mainstream muscle artifact removal schemes in the denoising area.One scheme adopts state-of-the-art BSS methods without EMG reference while the other utilizes auxiliary EMG channels as a reference in the denoising process.The above introduced SSA-CCA belongs to the first scheme.However,based on our previous studies,we clearly and deeply realize that how to more effectively remove the inevitable and complex EMG artifacts from EEG with only one channel or a few channels,is worth our further investigation.Steady-state visual evoked potential(SSVEP)serves as one of extensively utilized paradigms for brain-computer interface(BCI).SSVEP-based BCI has advantages of good classification accuracy,high information transfer rate(ITR)and little user training.Nowadays in the dynamic scenario,there exist few studies investigating wearable SSVEP-based BCI devices for muscle artifact contamination.Thus aimed at this problem,our study systematically explores the above two major muscle artifact removal schemes.The results of real collected SSVEP data demonstrate that under the mobile few-channel situation,it is of necessity to mount additional EMG channels for effective muscle activity suppression.This work proposes algorithms of innovation and verifies an effective scheme for muscle artifact removal from mobile EEG devices,applied to epilepsy data and wearable SSVEP-based BCI respectively.We believe that this study will provide timely and important guideline for the research direction of wearable EEG devices and will further inspire more relevant studies especially in the coming mobile,wearable and hybrid intelligent era.