Electroencephalography (EEG) is a non-invasive technique used to record electrical activity of brain using electrodes placed on the scalp. EEG signals always contain artifacts. Characteristics of EEG technique allow recording of brain activity under dynamic conditions. However, artifact contamination increases in frequency and complexity with movement intensity. Removal of artifacts from EEG signals recorded under dynamic conditions demands different approach in comparison to static conditions. Based on literature review we decided to compare methods AMICA and ASR and their combinations. We evaluated these methods on simulated data and on a example of real EEG signals recoded during human movement. We used 3D model of the head to simulated EEG signals with different noise intensity. Overall, artifacts were removed most thoroughly with combination of both methods. AMICA methods is good for removal of periodic artifacts, while ASR is good for removal of non-periodic artifacts with large amplitudes, therefore these two methods complement each other. Efficiency of artifact removal depends heavily on characteristics of signals and artifacts. In order to create useful guidelines about artefact removal in dynamic conditions further evaluation of methods for artifact removal on EEG signals recorded during specific types of movement and different intensities is needed.
|
