Fig 1.
ETDRS grid and example 10 intra-retinal layer segmentation.
(A) Standard ETDRS grid showing the foveal subfield (black). The inner ring is an average of the four parafoveal subfields (dark grey) and the outer ring of the four perifoveal subfields (light grey). (B) Screenshot of 10 layer (11 boundary) segmentation of a long-wavelength OCT image, produced by the Iowa Reference Algorithms. The left half of the image shows the image prior to segmentation. Layers 1–10 (top to bottom; as defined by the software): retinal nerve fiber layer (RNFL); ganglion cell layer (GCL); inner plexiform layer (IPL); inner nuclear layer (INL); outer plexiform layer (OPL); outer plexiform layer-Henle fiber layer to boundary of myoid and ellipsoid of inner segments (OPL-HFL ~ BMEIS); photoreceptor inner/outer segments (IS/OS); inner/outer segment junction to inner boundary of outer segment photoreceptor/retinal pigment epithelium complex (IS/OSJ ~ IB_RPE); outer segment photoreceptor/retinal pigment epithelium complex (OPR); retinal pigment epithelium (RPE).
Table 1.
Mean thickness of 10 intra-retinal layers.
Thickness values (mean ± SD; μm) produced by segmentation of images at session 1 using the Iowa Reference Algorithms.
Table 2.
Total retinal thickness measurements from different segmentation methods.
Thickness values (mean ± SD; μm) for the three ETDRS regions of images acquired at session 1.
Table 3.
Agreement of total retinal thickness between OCT devices.
Mean difference (bias; μm) and 95% limits of agreement (μm) for mean retinal thickness at session 1 produced by the Iowa Reference Algorithms, using AEL-dependent scaling, for each pairing of OCT instruments.
Fig 2.
Correlation of retinal thickness differences against AEL.
Difference in retinal thickness measurements (μm) versus axial eye length (mm) for the foveal subfield, as produced by the Iowa Reference Algorithms using two different transverse scaling methods.
Fig 3.
Intra-retinal layer thickness and CoR.
Mean thickness of 10 intra-retinal layers segmented by the Iowa Reference Algorithms at session 1, on images from all three OCT devices. Error bars and table values represent inter-session CoR (μm) for each layer.
Table 4.
Inter-session repeatability of the Iowa Reference Algorithms.
Coefficients of repeatability (μm; and percentage) of mean retinal thickness produced by the Iowa Reference Algorithms at session 1 and session 2.
Table 5.
Inter-session repeatability of the on-board software.
Coefficients of repeatability (μm; and percentage) of mean retinal thickness produced by the on-board analysis software of the commercial instruments at session 1 and session 2.
Fig 4.
Bland-Altman plots showing inter-session repeatability.
Total retinal thickness difference (μm) against mean (μm) for inter-session repeatability of the foveal subfield for both segmentation methods. 95% limits of agreement shown by dashed lines. Note that two significant outliers were removed from the Topcon on-board segmentation data.
Fig 5.
Comparison of inter-session repeatability for the two segmentation methods.
Coefficients of repeatability for segmentation by the Iowa Reference Algorithms and on-board software are shown for the Topcon (left) and Zeiss (right) systems. The ‘equivalent retinal thickness’ values from the Iowa Reference Algorithms were used in this analysis.
Table 6.
Agreement of total retinal thickness between segmentation methods.
Mean difference (bias; μm) and 95% limits of agreement (μm) for mean retinal thickness at session 1 produced by the Iowa Reference Algorithms and the commercial on-board equivalent.