Review of Motion Artifact Removal From an Electroencephalogram

‍ ‍A brain-computer interface （BCI） is a system that directly converts central nervous activity into artificial output without relying on peripheral nerves and muscles. An electroencephalogram （EEG） is an important means of BCI implementation. Because of its advantages of being non-invasive with a high time resolution， the EEG has become one of the most widely used brain information acquisition technologies. However， an EEG recorded on the scalp is transmitted and attenuated by the cranium and other media， the signal amplitude is weak， the signal-to-noise ratio is low， and it is easily interfered by physiological noise， such as ECG， ophthalmic electricity， myoelectric noise， and non-physiological noise， such as power frequency interference. Considering the miniaturization and portability of EEG acquisition equipment， the application range of EEGs has gradually expanded to more dynamic environments. In actual motion scenes in non-laboratory environments， collected EEG signals are usually mixed with a large number of motion artifacts. The motion artifact interference frequency is low， the amplitude is large， the harmonic range is wide， and the frequency range easily coincides with the brain frequency band. Therefore， it is difficult to eliminate or reduce motion artifacts. Motion artifacts can seriously affect the quality of EEG signals and hinder EEG decoding. In response to this problem， many researchers at home and abroad have carried out in-depth studies on the removal methods of motion artifacts in EEG signals， but there is a lack of systematic inductions and summaries of these studies. Therefore， this study focuses on the removal techniques of EEG motion artifacts. This study summarizes the main evaluation indexes of motion artifact removal； addresses two types of motion artifact removal algorithms， both without a reference and with gesture data as a reference； systematically compares the advantages and disadvantages of existing algorithms and their applicable scenarios； and predicts the development direction of motion artifact removal to promote the wide application of EEG technology in real scenes.