Abstract
Purpose :
OCT parameters commonly used for glaucoma diagnosis include circumpapillary retinal nerve fiber layer thickness (RNFLT) and optic nerve head (ONH) neuroretinal rim measures such as mimimum rim width (MRW) and area (MRA) derived from circular and radial B-scans, respectively. However, OCT angiography scans are most commonly based on a raster grid B-scan pattern. We sought to determine if RNFLT, MRW and MRA parameters derived from circular and radial B-scans sampled from the grid volume (virtual B-scans) are comparable to those derived directly.
Methods :
OCT scans (Spectralis OCT2, Heidelberg Engineering, GmbH) were obtained longitudinally in 18 eyes of 9 rhesus monkeys at baseline and though an early stage of unilateral experimental glaucoma (EG). The scan pattern used for direct derivation of MRW and MRA was comprised of 24-radial B-scans (15° span, 768 A-lines, averaged 25x, centered automatically on the ONH Bruch’s membrane opening, BMO). RNFLT was derived directly from a circular B-scan (6° radius, 1536 A-lines, averaged 100x). Grid scans contained 768x768 A-lines over 15°x15°, manually centered on the ONH, averaged 5x. Virtual B-scans were then sampled from the grid to create 24 15° radials and a circular 6° radius B-scan, all centered on the manually delineated BMO. Bland-Altman analysis was used to evaluate differences between parameters derived directly from standard recorded B-scans versus virtual B-scans.
Results :
The 95% limits-of-agreement (LOA) for the differences between parameters derived directly, versus from virtual B-scans, did not differ between control eyes and EG eyes (Fig1, Table 1). For all eyes combined, the 95% LOA were: -18.1 to 19.8 µm for MRW; -0.064 to 0.101 mm2 for MRA; -0.098 to 0.110 mm2 for BMO Area; and -7.6 to 8.2 µm for RNFLT, representing ± 5.9%, 7.6%, 8.3% and 7.7% of their mean values, respectively. These LOA are better than the limits of test-retest repeatability for MRW and MRA and similar to test-retest limits for BMO Area and RNFLT derived from standard B-scans.
Conclusions :
When they are derived from virtual B-scans sampled from a dense OCT grid pattern, the parameters MRW, MRA and RNFLT are comparable to those derived directly from specifically dedicated B-scan patterns. Our results show that these clinically diagnostic parameters can be derived from a single scan type, such as that used for OCT angiography.
This is a 2020 ARVO Annual Meeting abstract.