Visual field recovery after vision restoration therapy (VRT) is independent of eye movements: an eye tracker study

It has been argued that patients with visual field defects compensate for their deficit by making more frequent eye movements toward the hemianopic field and that visual field enlargements found after vision restoration therapy (VRT) may be an artefact of such eye movements. In order to determine if this was correct, we recorded eye movements in hemianopic subjects before and after VRT.

Visual fields were measured in subjects with homonymous visual field defects (n=15) caused by trauma, cerebral ischemia or haemorrhage (lesion age >6 months). Visual field charts were plotted using both high-resolution perimetry (HRP) and conventional perimetry before and after a 3-month period of VRT, with eye movements being recorded with a 2D-eye tracker. This permitted quantification of eye positions and measurements of deviation from fixation.

VRT lead to significant visual field enlargements as indicated by an increase of stimulus detection of 3.8% when tested using HRP and about 2.2% (OD) and 3.5% (OS) fewer misses with conventional perimetry. Eye movements were expressed as the standard deviations (S.D.) of the eye position recordings from fixation. Before VRT, the S.D. was +/-0.82 degrees horizontally and +/-1.16 degrees vertically; after VRT, it was +/-0.68 degrees and +/-1.39 degrees , respectively. A cluster analysis of the horizontal eye movements before VRT showed three types of subjects with (i) small (n=7), (ii) medium (n=7) or (iii) large fixation instability (n=1). Saccades were directed equally to the right or the left side; i.e., with no preference toward the blind hemifield. After VRT, many subjects showed a smaller variability of horizontal eye movements. Before VRT, 81.6% of the recorded eye positions were found within a range of 1 degrees horizontally from fixation, whereas after VRT, 88.3% were within that range. In the 2 degrees range, we found 94.8% before and 98.9% after VRT. Subjects moved their eyes 5 degrees or more 0.3% of the time before VRT versus 0.1% after VRT. Thus, in this study, subjects with homonymous visual field defects who were attempting to fixate a central target while their fields were being plotted, typically showed brief horizontal shifts with no preference toward or away from the blind hemifield. These eye movements were usually less than 1 degrees from fixation. Large saccades toward the blind field after VRT were very rare.

VRT has no effect on either the direction or the amplitude of horizontal eye movements during visual field testing. These results argue against the theory that the visual field enlargements are artefacts induced by eye movements.