Purchase this article with an account.
Steven Seto, Alison L Huckenpahler, Emma Warr, Alexander E Salmon, Joseph Carroll; Repeatability and Reproducibility of Ellipsoid Zone Intensity Measurements on SD-OCT Images. Invest. Ophthalmol. Vis. Sci. 2019;60(9):1289. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The integrity of the ellipsoid zone (EZ) on OCT images is used as a measure of photoreceptor structure in a variety of retinal diseases. Central to advancing the EZ as a photoreceptor biomarker is the development of robust methods for quantifying EZ reflectivity. Here we assess intra-observer repeatability and intra- and inter-observer reproducibility of a method to measure EZ intensity.
Monocular horizontal 7mm line scans were acquired through the fovea using the Bioptigen OCT (1000 A-scans/B-scan and 80 B-scans) in 14 individuals with no known ocular disease. Each line scan was processed by registering and averaging 20 frames using the TurboReg plugin in ImageJ. The resultant images were flattened to the RPE contour and interpolated by 10x for analysis. Images were analyzed twice by 3 observers of varying proficiency as follows: The fovea was located subjectively using the peak outer segment length, then longitudinal reflectivity profiles (LRPs) were sampled at 0.5mm intervals out to 2.5mm nasally and temporally – generating 10 LRPs per image. For each LRP, the peak intensity of the EZ was extracted. EZ intensity was normalized to the average intensity of the IPL for that LRP. Intra-observer repeatability was estimated using the repeatability coefficients derived from within-subject standard deviation. Six subjects returned for repeat imaging 1 week later and were analyzed by a single observer to examine the intra-observer reproducibility, assessed using within-subject standard deviation.
There was no significant difference in EZ intensity measurements between the 3 observers (p=0.18, Friedman test). The repeatability coefficient was comparable across the 3 observers of varying levels of experience. The repeatability coefficients were 0.52 (12.6%), 0.45 (11.0%), and 0.41 (10.0%) for the expert, trained, and naïve observer, respectively. Thus, the difference between 2 measurements of EZ intensity from the same scan would be expected to be less than about 11% for 95% of pairs of observations. The intra-observer reproducibility was worse when comparing images taken 1 week apart (1.12; 34.1%).
Measurements of EZ intensity can be obtained with high repeatability regardless of observer proficiency. However, measurements made over time showed worse reproducibility, which may affect longitudinal examinations of EZ reflectivity as a potential marker of disease progression.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.
This PDF is available to Subscribers Only