Performance of neuroretinal rim thickness measurement by Cirrus high-definition optical coherence tomography in myopic eyes

Abstract

Neuroretinal rim (NRR) measurement can aid the diagnosis of glaucoma. A few studies reported that Cirrus optical coherence tomography (OCT) had NRR segmentation errors. The current study investigated segmentation success of NRR in myopic eyes using the Cirrus built-in software and to determine the number of acquisitions required to identify NRR thinning. Right eye of 87 healthy adult myopes had an optic disc scanned using Cirrus HD-OCT for five successive acquisitions. A masked examiner evaluated 36 radial line images of each scan to screen for segmentation errors using the built-in software at the Bruch’s membrane opening (BMO) and/or internal limiting membrane (ILM). Participants with three accurate NRR acquisitions had their average NRR thickness determined. This result was compared with average of the two acquisitions and the first acquisition. Among 435 OCT scans of the optic disc (87 eyes × 5 acquisitions), 129 (29.7%) scans had segmentation errors that occurred mainly at the ILM. The inferior-temporal and superior meridians had slightly more segmentation errors than other meridians, independent of axial length, amount of myopia, or presence of peripapillary atrophy. Sixty-five eyes (74.7%) had at least three accurate NRR measurements. The three acquisitions had high reliability in NRR thickness in the four quadrants (intraclass correlation coefficient > 0.990, coefficient of variation < 3.9%). NRR difference between the first acquisition and the average of three acquisitions was small (mean difference 2 ± 13 μm, 95% limits of agreement within ± 30 μm) among the four quadrants. Segmentation errors in NRR measurements appeared regardless of axial length, amount of myopia, or presence of peripapillary atrophy. Cirrus segmentation lines should be manually inspected when measuring NRR thickness.