Abstract
Purpose :
To evaluate the inter-photographer, inter-session, intra-grader and inter-grader reliabilities of Adaptive Optics Scanning Laser Ophthalmoscopy (AOSLO) for cone density estimation in healthy human eyes in a simulated real-life clinical setting.
Methods :
We performed a prospective study of ten eyes without any eye disease (mean age of 30.2 years, 60% female). Macular cones were imaged in two 680 × 680 µ2 squares using an AOSLO camera (AOSLO RD102, Canon, Tokyo, Japan) in a simulated clinical setting with clinic staff photographers.The first square was centered on the fovea and the second one was nasal to the first square.Two masked ophthalmologist-graders measured cone counts in 150 × 150 µ2, 80 x 80 µ2, and 50 X 50 µ2 squares within the imaged region. Image-to-image square localization was determined by the graders using a reference fundus photograph because the camera lacks tracking software. Inter-grader and intra-grader reliabilities were determined using the highest quality images. Intra-grader agreement was evaluated by comparing identical frames measured on two separate occasions by the same grader. Inter-session repeatability was assessed by five separate instances of the photographer focusing the camera onto the same region.To determine inter-photographer reproducibility, each patient was imaged by at least two photographers. Statistical analyses were performed by calculating intraclass correlation coefficients (ICC) with Stata (College Station, Texas).
Results :
Cone cell density measurements were most consistent with the 150 x 150 µ2 squares. Intra-grader and inter-grader ICCs were 0.99 [95% confidence interval (CI) = 0.997 – 0.999] and 0.98 (95% CI = 0.975 – 0.989), respectively. Inter-session repeatability had an ICC = 0.72 (95% CI = 0.423 – 0.909). Inter-photographer reproducibility had an ICC = 0.72 (95% CI = 0.461 – 0.866). One common reason for discordance in cone density measurements between images taken by different photographers was variability in cone imaging quality. Image quality also affected the inter-session repeatability.
Conclusions :
In a simulated real-life clinical setting, AOSLO had excellent inter-grader and intra-grader reliability for healthy eyes. Inter-session and inter-photographer reliabilities were not as good but still strong. Variations in image quality strongly impacted cone analysis reliability.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.