Reliability of Counting Cone Inner Segments in Choroideremia

Abstract

Purpose : Recent advances in adaptive optics scanning light ophthalmoscopy (AOSLO) have allowed the visualization of cone inner segments (IS) and enabled quantification of IS cone density in normal and diseased retina (Scoles, et al. IOVS, 2014). Understanding the reliability of IS cone density measurements is essential for using changes in IS cone density as an outcome measure in longitudinal studies of disease progression and assessment of experimental intervention. Here, we examine the inter-observer reliability for measuring IS cone density in patients with Choroideremia.

Methods : The central two mm of 11 eyes from 11 Choroideremia subjects age 9 to 37 were imaged along the horizontal and vertical meridians using nonconfocal split detection AOSLO. AOSLO images were semi-automatically montaged using a custom MATLAB algorithm (Chen, et al. Biomed. Optics Ex., 2016). Regions of interest (ROIs) containing cone IS were selected at 190, 350, 500, and 900 μm in all four directions from the fovea. An ROI was included for density analysis if the image fell within the patient’s central island of intact retina but excluded if the ROI fell outside of the atrophic border, resulting in 136 ROIs. Three trained observers (previously shown to exhibit high agreement in normal images, Han G, et al. IOVS 2016;57:ARVO E-abstract 4640) manually identified the cone IS using custom software. Cone density was extracted from each ROI, and intraclass correlation coefficients with 95% confidence intervals (CI) were calculated to assess the inter-observer agreement in cone density measurements.

Results : Cone IS density was consistent with our previously reported cone densities in Choroideremia. Inter-observer agreement for cone density measurements was good at 350 and 500 μm (0.684, CI 0.526-0.810 and 0.741, CI 0.0.594-0.851) and moderate at 190 and 900 μm (0.543, CI 0.372-0.696 and 0.527, CI 0.283-0.739). Overall inter-observer agreement in cone density was lower for Choroideremia images than for normal images.

Conclusions : The dependence on inter-observer agreement with retinal eccentricity is consistent with the expected difficulty associated with identifying cone IS at different retinal locations in Choroideremia. Cone identification appeared consistent within observers, however further study of intra-observer repeatability is required. These studies will be important for using cone density as an outcome measure to assess disease progression in Choroideremia.

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.