Abstract
Purpose :
Adaptive optics images reveal foveal defects in the cone mosaic, with localized areas of dark cones clustered in the fovea and consistent with early edematous changes. Our previous result of worse fixation stability in patients with clinically significant macular edema was with wide-field imaging. Here we characterize fixation on a fine scale at the high resolution provided by adaptive optics imaging. We compared sample to sample displacements and velocities to the spatio-temporal characteristics known to provide optimal contrast sensitivity, i.e. non-zero velocities no faster than slow drift.
Methods :
We performed a new analysis on retinal images from an Indiana adaptive optical scanning laser ophthalmoscope, on series of 100 images collected at 30 Hz and aligned strip-wise to a template with 55 overlapping samples. This sampled retinal position up to 1650 Hz. The digital resolution is 1 micron/pixel, with sub-pixel registration and 6 steps horizontally and vertically, giving 2 arc sec digital resolution. The initial output included position with respect to the template, horizontal displacement, vertical displacement, and cross correlation quality. Next, we computed the sample to sample Euclidean distance from 85 aligned images, selected from a series in the first 5 blocks to minimize the effects of dry eye or fatigue. Both samples had a correlation coefficient < .75. We used the first 2052 computations.
Results :
For diabetic patients with localized areas of dark cones (N = 12. 50% F, 50% M, age = 49.8 + 12.3 yr), the frequency distributions of displacements were skewed towards small displacements. An average of 93.4 + 6.79% of the displacements were < 1 micron. The displacements < 1.4 microns (0 to 15 cycles/deg/sec) were also skewed to smaller displacements: 14% occurred at 1.41 + .47 cycles/deg/sec but significantly larger than the displacement category containing 0 (p < .01). Alignment errors due to large motions, blinks, or optical artifacts occurred in a low proportion of samples: 88.9 + 27.2%.
Conclusions :
Using the alignment of strips from retinal images acquired with an AOSLO, we achieved eye motion sampling of 2 arc sec at 1650 Hz for 89% of the samples, even in the eyes of diabetic patients with early signs of edema. Gaze was dominated by small, low velocity eye movements that could improve spatial contrast sensitivity.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.