Abstract
Purpose :
The human cone mosaic can be imaged noninvasively with adaptive optics scanning light ophthalmoscopy (AOSLO), allowing direct measurements of foveal cone density. However, the presence of severe nystagmus or dense cataract precludes obtaining images with adequate quality to make these measurements. Optical coherence tomography (OCT), however, is less affected by these issues and allows measurement of outer segment (OS) length. Here, we examined the relationship between foveal cone density and OS length to determine if OCT could be used to estimate peak cone density.
Methods :
Forty-seven subjects (23 normal, 21 albinism, 3 premature birth; aged 6-67 years) were imaged with OCT and AOSLO. OCT images were averaged and manually segmented; foveal OS length was taken as the maximum distance between the 2nd and 3rd hyper-reflective bands. AOSLO images were registered and averaged to create a foveal montage, from which cones were semi-automatically identified to derive an estimate of peak cone density. We modeled the expected relationship between peak cone density and OS length with the following assumptions: that the cylindrical OS has constant volume across subjects (141 μm3), a ratio of cone OS width to inner segment width of 0.73, and that the cone mosaic is arranged in a perfect crystalline lattice.
Results :
OS length ranged from 24.50 to 53.82 microns (avg ± stdev = 38.81 ± 8.50 μm). Peak cone density ranged from 29,218 to 214,020 cones/mm2 (113,559 ± 45,266 cones/mm2). Peak cone density and OS length were significantly correlated (Spearman r = 0.79; p < 0.0001). While our measurements followed the general relationship predicted by the model (i.e., foveas with longer OS had higher peak cone density), there is a discrepancy between the data and model that is likely due to deficiencies in the current model.
Conclusions :
In retinas with contiguous foveal cone mosaics, OS length and peak cone density are significantly correlated. This suggests that OS length could provide reasonable estimates of peak cone density in these subjects when AOSLO images are not available. However, this approach would not be suitable for subjects with non-contiguous foveal cone mosaics. Further refinement of the anatomical model employed here is needed, and may yield important insight into aspects of foveal cone specialization.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.