Removing “cleaned” eye-blinking artifacts from EEG measurements

Electroencephalography (EEG) is a useful tool for brain research. However, during recordings, many physiological or technical artifacts can be observed. Such artifacts might hide the brain information and should be removed. In this paper, we aim at suppressing the ocular artifacts by combining Independent Component Analysis (ICA) and an adaptive Wiener filter ("ICA-WF"). The idea is to obtain pure eye blinking components and suppress them from the original independent components in order to remove the artifacts and preserve the physiological information. A comparison between three methods to suppress eye blinking artifacts in EEG signals is also presented: ICA with removing completely the artifactual components ("ICA-Complete"); ICA with removing only the contaminated segments of the artifactual components ("ICA-Partial"); and the proposed combination of ICA with a Wiener filter. These methods are applied to real EEG data from a healthy subject. Both by visual inspection and in a quantitative manner, it is demonstrated that ICA-WF suppresses the eye blinking artifacts much more efficiently than the other two methods.