1 Biomedical Engineering, Marquette University, Milwaukee, WI
2 Ophthalmology, Medical College of Wisconsin, Milwaukee, WI
3 New York Eye and Ear Infirmary, New York, NY
4 Mount Sinai School of Medicine, Mount Sinai Hospital, New York, NY
5 The Institute of Optics, University of Rochester, Rochester, NY
6 Biomedical Engineering, University of Rochester, Rochester, NY
7 Biophysics, Medical College of Wisconsin, Milwaukee, WI
8 Cell Biology, Medical College of Wisconsin, Milwaukee, WI
Commercial Relationships: Robert Cooper, None; Zachary Harvey, None; Michael Dubow, None; Yusufu Sulai, None; Alexander Pinhas, None; Drew Scoles, None; Nishit Shah, None; Richard Rosen, Opko-OTI (C), Optos (C), Clarity (C), OD-OS (C), Topcon (R), Zeavision (F), Genetech (F), Optovue (C); Alfredo Dubra, US Patent No: 8,226,236 (P); Joseph Carroll, Imagine Eyes, Inc. (S)
Purpose:Adaptive Optics Scanning Light Ophthalmoscopes (AOSLOs) permit near diffraction-limited imaging of the human photoreceptor mosaic, though intraframe eye movements lead to image distortion. Here, we explore the impact of these distortions on a number of metrics commonly used to characterize the photoreceptor mosaic.
Methods:We acquired 9 image sequences of the parafoveal cone mosaic from 7 subjects on 3 similar AOSLOs. In another subject, we acquired an AOSLO and flood-illuminated AO image sequences of the same retinal location. To assess the effect of distortions within AOSLO images, ten averaged images were produced by registering against different reference frames using a previously described algorithm. The images were then registered with the same software while tracking the distortion applied to each image. The photoreceptor coordinates from the reference frame were transformed using this distortion. Voronoi geometry, cone density, nearest-neighbor distance (NND), inter-cell spacing (ICS), and regularity index (RI) were calculated for each set of images. Repeatability was calculated to assess the effect of intraframe distortion on these metrics.
Results:Across the AOSLO images, we analyzed 17,942 cones, 75% of which retained the number of sides in the corresponding Voronoi domains across the 10 images (range 56%-90%). Cone density was found to have a repeatability of 1.8% (i.e., the difference between any 2 measurements on the same subject would be less than 1.8% for 95% of observations). NND and ICS had even better repeatability, at 1.4% and 0.95%, respectively. In contrast, the NND RI and ICS RI had a repeatability of 11% and 31%, respectively. Comparing an AOSLO image set to a flood-illuminated AO image, we found similar repeatability (density: 2.7%, NND: 0.7%, ICS: 0.83%, NND RI: 8.9%, ICS RI: 20.3%) and 83% of cells had retained Voronoi geometry.
Conclusions:Global metrics (density and cell spacing) are minimally affected by intraframe distortions, whereas local metrics (regularity index and Voronoi geometry) are more significantly affected. Intraframe distortion in AO scanning instruments limits the measurement accuracy of mosaic geometry, thus every effort should be made to choose minimally distorted reference frames.
Keywords: 549 image processing 648 photoreceptors 522 eye movements
© 2013, The Association for Research in Vision
and Ophthalmology, Inc., all rights reserved. Permission to republish any
abstract or part of an abstract in any form must be obtained in writing
from the ARVO Office prior to publication.