Dipole Analysis of Eye Movement Artifacts from the EEG

Independent Component Analysis (ICA) is an efficient technique for blind source separation. With dipole model, the sources can be localized and modeled as dipoles whose parameters may be computed based on the observed scalp EEG. In this study, We evaluated the use of Independent component analysis combining EEG dipole model to automatically recognize eye movement artifacts from EEG without the reference to EOG. We separated the EEG data into independent components using the ICA method, and determined the source localization of these independent components with a single dipole model. EEGs from 12 patients were analyzed. The experimental results indicate that ICA with dipole model is very efficient at recognizing the eye movement artifacts. The proposed method by combing Independent Component Analysis with Dipole Model is a generally applicable and effective method for recognizing ocular artifacts from EEG recordings, although slow waves and ocular artifacts share similar frequency distributions.