Agreement and precision analyses of retinal measurements using Swept-Source and Spectral-Domain OCTs

Abstract

Purpose : To evaluate and compare the precision and agreement of macular and optic disc measurements between Swept-Source and Spectral-Domain OCTs in normal and diseased eyes.

Methods : All subjects underwent OCT imaging using DRI OCT Triton (SS-OCT) and 3D OCT-1 Maestro (SD-OCT) [Topcon Corp., Tokyo, Japan] in three default scan modes: Wide (12x9 mm²), Macula (7x7 and 6x6 mm² for Triton and Maestro, respectively), and Disc (6x6 mm²). Six devices (3 Tritons and 3 Maestros) were operated by 3 photographers (each was paired with a Triton and a Maestro). 76 subjects (mean age, 59.3 ± 16.8 years), consisting of normal (n=25), retinal (n=26), and glaucoma (n=25) eyes were enrolled. In a single session, each study eye was imaged with at least three scans of each scan mode by all photographers/devices in a randomized fashion. Agreement and precision analyses were performed in measurements for macular full retinal thickness, ganglion plus inner plexiform layer thickness, ganglion cell complex thickness, circumpapillary retinal nerve fiber layer thickness, and optic disc parameters. Image quality of all scans was checked and only the acceptable scans were included in analyses. All acceptable scans were entered into a two way random-effect ANOVA model for precision analysis. For agreement, the first acceptable scan of each study eye was used for Deming regression and Bland-Altman analysis.

Results : Repeatability and reproducibility between Triton and Maestro were comparable across all parameters, while the Triton scan modes overall have more favorable precision results. Repeatability limit and reproducibility limit ratios (Triton/Maestro) ranged from 0.14 – 1.64 and 0.24 – 1.22, respectively, among all parameters. There were small differences (Triton – Maestro) among various thickness measurements (ranged between -5.49 – 0.90 micron), though these measurements were not clinically significant as they were around or below the digital and axial resolutions, as well as less than the Maestro’s reproducibility limits. Some differences in optic disc measurements were reported. These were the result of improved optic disc detection in the Triton.

Conclusions : Overall, the precision and agreement results indicated good agreement between Triton and Maestro. Optic disc detection improvement in Triton produced subtly more reasonable disc margin delineations.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.