Abstract
Purpose:
To evaluate a commercially available flood-illuminated adaptive optics camera for its repeatability in measuring the cone mosaic and to determine average cone density among healthy subjects
Methods:
We used the RTX1 flood-illuminated adaptive optics camera (Imagine Eyes: Orsay, France) to image the cone mosaic in 54 healthy subjects, ages 14 to 69 years. For each subject, a series of 25 4°x4° retinal images were obtained. Using MosaicJ, these images were combined to create a retinal montage spanning a 12°x12° field of the central macula. Retinal montages were analyzed for regional differences in cone density. Images were post processed in ImageJ and cone counting was performed with a MATLAB algorithm. To determine repeatability of the device and montaging process, a subset of 7 subjects ranging in age from 19 to 36 years were imaged on 3 separate occasions. To assess the validity of automated cone counting for each montage, automated cone counts were compared to manual counts in sample retinal areas 2° and 4° temporal to the foveal center.
Results:
Image quality was excellent in most young subjects, but increasingly variable in older subjects. Cones within 1.5° of the foveal center could not be resolved with this camera. Angular density of cones averaged 1587 ± 91 cones/degree2 at 1.6° - 4.3° eccentricity and 1428 ± 64 cones/degree2 between 4.3° and 5.4° eccentricity. When adjusted for axial length, cone density averaged 18,688 ± 2081 cones/mm2 between 1.6° and 4.3° and 16,763 ± 1669 cones/mm2 between 4.3° and 5.4°. Angular cone density between 1.6° and 4.3° decreased with age, but between 4.3° and 5.4° increased. Repeated measurements of cone density in the same subject from separate imaging sessions resulted in an intraclass correlation coefficient of 0.98 (p < 0.001, 95% CI: 0.93-1.00) between 1.6° and 4.3° and 0.95 (p < 0.001, 95% CI: 0.83-0.99) between 4.3° and 5.4°. However, validity of the automated cone counting algorithm depended on image quality, with poor quality images producing higher variability.
Conclusions:
Flood-illuminated adaptive optics provides measurements of cone density that are consistent with anatomical studies. Repeated measurements in a subset of younger subjects resulted in very strong intraclass correction coefficients, which indicate the system produced consistent measurements. However, challenges remain with older subjects.
Keywords: 550 imaging/image analysis: clinical