Abstract
Purpose :
The purpose of this study was to determine the topographical variability in determination of macular layer thickness by automatic layer segmentation in two models of SD-OCT instruments during two visits 2-3 weeks apart. As a continuation of previous work, this study focused mostly on outer retina layers.
Methods :
SD-OCT scans in Posterior Pole mode (8 x 8 grid; 64 cell areas) centered on the fovea were performed on normal volunteers. Three consecutive images of each eye were acquired on two models of Spectralis SD-OCT (Heidelberg Engineering): 2011 model (OCT1) operating at 40 KHz scanning rate and a newer 2017 model (OCT2), operating at 85 KHz scanning rate. The imaging was repeated in 2-3 weeks. Four retinal layers: inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear layer (ONL) and retinal pigment epithelial layer (RPE) were segmented automatically by the Spectralis software. For each grid cell the root mean square (RMS) deviation from the mean of each set of three measurements was calculated in two ways: as an absolute error (in microns) and as a relative error (as % of corresponding layer thickness). Heatmaps of error distribution within the imaged grid were generated and analyzed to determine the topographical distribution of error.
Results :
Both eyes of 14 volunteers (4 men and 10 women) aged 35.1 +/- 12.3 years were imaged. The average relative RMS error was distributed differently for some layers, but similarly between eyes, instruments, and showed similar pattern between the two visits. Thus, the error was more than 2 times larger in the central 4 grid cells (fovea and parafovea) for OPL and ONL compared to the rest of the cells (ratio central vs. peripheral 2.30 +/-0.28 and 2.28+/-0.23, respectively). This ratio was significantly larger compared to the corresponding ratio for either INL or RPE (0.97+/-0.09, 1.14+/-0.22, p<0.001, paired t-test). Conversely, the error was generally larger in nasal periphery for INL (ratio most nasal column of cells vs rest 1.39+/- 0.16, p<0.0001, one sample t-test), while there was no obvious pattern of error distribution for the RPE.
Conclusions :
Relative RMS error was not uniformly distributed across the imaging area for all layers. These findings would help evaluation of the changes in macular thickness for clinical purposes.
This is a 2020 Imaging in the Eye Conference abstract.