The Study On Motion Artifact Removal From EEG Of Portable Devices

Electroencephalogram(EEG)refers to the signals generated by neural activity in the brain,which were recorded by sensors close to the scalp through volume conduction of electrical signals.EEG is a non-invasive brain electrical signal acquisition method that can reflect the brain activity process in real time with high time resolution.It has been widely used in scientific research and medical field.The EEG signal is very weak,and it is easy to be contaminated by various artifacts during the acquisition process.Therefore,EEG is usually collected in a clinical or laboratory environment,and medical staff or researchers take additional precautions to avoid recording artifacts.However,with the promotion of portable EEG acquisition equipment,when EEG signal acquisition is transferred from hospitals or laboratories to homes,motion artifacts will seriously interfere with the quality of EEG.Although some studies have proposed various methods to remove motion artifacts to obtain high-quality EEG signals,most of these studies are either limited to a highly controlled laboratory environment,or the effect is not very satisfactory.In this paper,we combine EEMD and PCA to propose an EEMD-PCA method to remove motion artifacts in the EEG collected by portable EEG devices.EEMD-PCA is a single-channel motion artifact removal method.First,we use EEMD to decompose a single-channel EEG into multiple intrinsic mode functions;Then we use PCA to separate multiple intrinsic mode functions into source components;Finally,the autocorrelation of the sources are used to remove the artifact components,and the signal is reconstructed to obtain the EEG without motion artifacts.In order to verify the effectiveness of this method,we designed an experiment,using the Emotiv Epoc portable EEG helmet to collect EEG data containing motion artifacts.Then it is tested on the simulated EEG data set and the real EEG data set.At the same time,it is compared with the two commonly used single-channel artifacs removal methods in EEG,which are EEMD-ICA and EEMD-CCA methods.For simulated data,compared with the current best EEMD-CCA method,the EEMD-PCA method has reduced the RMSE by 10.3%,increased the SNR by 3.1d B,and increased the correlation coefficient by 7%.The effect has been significantly improved.For real data,although the 14 channels are contaminated by motion artifacts to different degrees,the EEMD-PCA method can be effectively removed,showing good adaptability.Finally,the results show that the EEMD-PCA method performs best both in the simulated data and the real data.Therefore,EEMD-PCA is an effective tool that can remove motion artifacts from EEG collected by portable EEG devices.