Purpose
We aimed to test the hypothesis that automated eye-movement perimetry can yield visual threshold estimates that are similar to threshold estimates obtained using automated manual perimetry.
Methods
We tested 27 normal subjects with two perimetry conditions: manual, in which responses were made with a computer mouse; and eye-movement, in which responses were made by moving the eyes to the stimulus. Eye movements (monocular, right eye) were recorded with an EyeLink1000 infrared camera. System accuracy measuring the horizontal visual field was <0.5°, with a temporal resolution of 1000 Hz. Our analysis included responses made to 3 stimulus positions spread across the visual field (3°,3°; 3°,-9°; and -21°,-3°), and we used correlation measures to compare visual threshold data from the manual and eye-movement tasks for each stimulus position within participants. We manipulated target salience by varying stimulus size across a continuous range.
Results
The accompanying figure shows estimates of visual threshold obtained from manual perimetry (ordinate) against estimates obtained from eye-movement perimetry (abscissa) for each test location (one per panel). A strong and statistically significant positive correlation between manual and eye-movement thresholds was observed at each test location (3°,3°: r = 0.94; 3°,-9°: r = 0.70; and -21°,-3°: r = 0.76; each p < 0.001). On average, threshold estimates based on manual perimetry were numerically smaller than estimates based on eye-movement perimetry (mean differences [SD] at 3°,3°: 0.16 dB [0.41]; 3°,-9°: 0.21 dB [0.64]; and -21°,-3°: 0.03 dB [0.46]), but this difference was only reliable at one target location (3°,-9°: T(21) = 4.45, p < 0.001). Furthermore, all differences were within the expected range of test-retest reliability for perimetry testing (i.e., 1 dB).
Conclusions
Manual and eye-movement perimetry produced similar estimates of visual threshold in normal subjects, supporting our hypothesis that automated eye-movement perimetry can provide a valid measure of visual function. Further refinement of this technique will provide a tool permitting automated, whole-field perimetry testing, along with novel, eye-movement based measures (eye movement latency and accuracy) that reflect the health of the visual system.
Keywords: 525 eye movements: saccades and pursuits •
758 visual fields •
642 perimetry