Is there any way to remove ocular artifacts from an EEG signal, other than applying independent component analysis, or manual rejection?
According to a review by Croft & Barry (2000), a number of possible methods exist to prevent, or reject ocular artifacts in the EEG;
Prevention/reduction of artifact instances
- Recording with eyes closed -
A very effective method, but eye closure alters the EEG, e.g., it increases alpha band activity;
- Let subjects fixate on a target -
Method reduces blinks, but does not eliminate them.
- Exclude trials with amplitudes larger than, e.g., 50 microV -
Method works only with a trial-based approach; artifacts below criterion level will remain.
- Subtract a certain standard voltage when an artifact occurs -
A crude method;
- Online attenuation of the EEG signal when an artifact is detected -
Works better as the former technique, but is still crude;
- Regression techniques where different electrodes are used to estimate the artifact component in other electrodes -
A more sophisticated method with apparently good results.
For details I refer to Croft & Barry (2000) and the primary literature cited therein.
- Croft & Barry Neurophysiol Clin (2000); 30: 5-19