Purpose
To rapidly estimate cone spacing properties of the normal cone photoreceptor mosaic and to measure local anisotropies in the hexagonal pattern using a cone-averaging method.
Methods
The Indiana high-resolution Adaptive Optics Scanning Laser Ophthalmoscope was used to image the cone photoreceptors of 5 normal healthy subjects (refractive error: -2.25±1.35 D, 28.8 ± 3.4 years old, 0.5% tropicamide dilated pupil).<br /> Measurement of cones were recorded while the subjects looked at each corner and the center of a 1 degree imaging field (0.5 Airy disk confocal aperture). In addition four strips of 2x5 degrees of cones corresponding to the four primary meridians (Temporal (T), Nasal (N), Superior (S), Inferior (I)) were recorded using a 2degrees imaging field (and 1.5 Airy disk confocal aperture).<br /> Montages of average images were generated using an automated algorithm (combining Matlab, i2k Retina and Adobe Photoshop). Cone spacing properties were analyzed using a custom program that automatically identified individual cones, within a window (50 or 100 microns) that varied with retinal location. Within each window, interior subregions around each cone were extracted and averaged, providing a “kernel” image of an average cone and its surrounding retina. From each kernel image, we measured the averaged cone spacing -computed as the first maximum of the radial profile, then estimated the orientation and spacing anisotropy of the hexagonal patterns based in determining the principle axes of the packing (fig.1).
Results
There was a lower averaged cone spacing (higher cone density) along the horizontal (T, N) meridians than along the vertical (S, I) meridians. Locally the cone spacing was lower in vertical than in horizontal axes for all meridians near the fovea (<1°) with horizontal/vertical ratio of: 1.08±0.04 (T), 1.04±0.05 (N), 1.03±0.03 (S) and 1.05±0.01 (I). This tendency was maintained for Temporal and Nasal meridians in the parafovea (up to 5°) while it reversed for Superior meridian (1.11±0.07 (T), 1.10±0.06 (N), 0.97±0.05 (S), 1.00±0.03 (I)).
Conclusions
The method allows rapid automated estimates of cone packing properties and provides an analysis of individual difference in cone spacing and local anisotropies of the hexagonal cone array.