UBC Theses and Dissertations
Removal of ocular artifact from visual evoked response recordings O’Toole, Dennis Michael
Potentials generated by the eye cause unwanted artifact in Visual Evoked Response (VER) recordings. These artifacts often contaminate the data in a systematic way that can lead to spurious experimental results. Although it is widely agreed that ocular artifact must be accounted for, the methods used to deal with this problem are varied. The present study compared four methods used to control ocular artifact; blink rejection, eyes closed, subtraction and regression. Twenty normal, female subjects were tested twice within the same session. Subjects watched light flashes of 4 intensities; 2, 30, 80, and 240 ft lamberts. The lights were presented at 1 hertz, reached maximum brightness in 0.5 msec and lasted for 0.5 sec. During testing the VER, and electroocculographic (EOG) response generated by a blink, were recorded. In the blink rejection method, any VER epoch that contained blink artifact was excluded from the average. The eyes closed method consisted of having subjects watch the stimuli through closed eyelids. The subtraction method corrects blink artifact by digitally subtracting the averaged EOG from the EEG. The proportion of EOG subtracted was determined by the EEG/EOG ratio estimated while subjects blinked spontaneously in a darkened environment. The regression method determines what proportion of EOG is to be subtracted on the basis of the correlation between EOG and EEG within VER epochs. Two correction, factors are calculated, one to correct for vertical movements and one to correct for horizontal movements. The blink rejection method was found to be useful with subjects who had 40% or more blink-free epochs, but was an unreliable method for the majority of subjects. The eyes closed method was also found to produce poor VER data. The eyelids appear to attenuate the light reaching the retina and there may be eyeball movement despite having the eyes closed. Both the subtraction and regression methods substantially reduced the ocular artifact. Horizontal eye movements do not appear to be a significant problem over the short intervals of VER recording because the regression method was not superior to the subtraction method in removing artifact. Although the subtraction and regression methods effectively reduce ocular artifact, both are less effective at posterior electrode placements. The reason for this may be that ocular potential is not propagated across the scalp in a linear fashion, as often assumed. Using spontaneously generated blinks in a darkened environment, it was found that the ocular potential waveform changes shape as it moves towards the back of the head.