Abstract
Purpose :
To investigate the repeatability of macular, peripapillary retina and optic nerve head (ONH) parameters using spectral domain optical coherence tomography (OCT) in patients with advanced glaucoma.
Methods :
Twenty-three eyes of 23 patients with advanced open-angle glaucoma (median age, 60.5 y, range 47 to 82 y and median visual field Mean Deviation, -26.7 dB, range -31.9 to -15.3 dB) had five Spectralis OCT images of the macula (61 horizontal scans), ONH (24 radial scans) and peripapillary retina (1 circular scan) at the same visit, made by one examiner. Only eyes with visual acuity of 0.30 logMAR or better were included. All acquisitions were made relative to the axis connecting the fovea to the Bruch’s membrane opening (BMO) center with the follow-up mode turned on. One examiner reviewed all B-scans, and segmentation errors were manually corrected when necessary. The within subject standard deviations (SW) of the ganglion cell layer (GCL) thickness,retinal nerve fiber layer (RNFL) thickness, and minimum rim width (MRW) were used to estimate the repeatability.
Results :
7015 macular scans, 115 peripapillary retina scans and 2760 ONH scans were reviewed. Only 2% of the macular scans had inaccurate segmentations of the GCL boundaries, while 32% of the peripapillary circular scans presented inaccurate identification of the RNFL limits. Of the radial ONH scans, 16% and 17% of the radial ONH scans presented segmentation errors in the BMO and the internal limiting membrane, respectively. Mean (SD) global GCL thickness, RNFL thickness and MRW were 21.4 (2.8) µm, 42.5 (6.0) µm and 103.9 (34.3) µm, respectively. The SWs (95% confidence intervals) for global GCL thickness, RNFL thickness and MRW were 0.35 (0.30, 0.40) µm, 0.68 (0.58, 0.78) µm and 1.75 (1.49, 2.00) µm, respectively.
Conclusions :
We found high repeatability of OCT parameters in patients with advanced glaucoma. The automated segmentation algorithm was able to identify layer boundaries with good reproducibility and few segmentation artifacts in most eyes.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.