Purchase this article with an account.
Enrico Pellegrini, Alan Fleming, Michael Sinai, Nikolas London, Jano van Hemert; Correlation of retinal thickness measures and optic nerve head parameters between different OCT scan types. Invest. Ophthalmol. Vis. Sci. 2020;61(9):PB00102.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Retinal imaging devices on the market can produce optical coherence tomography (OCT) scan patterns with different fields of view and resolution to depict different regions of the retina. It is important that quantitative measures of the same structure, such as macular and peripapillary thickness values or optic nerve head (ONH) parameters, remain consistent regardless of the scan pattern that they are measured from. We have retrospectively analyzed the correlation between retinal measures obtained by automated processing of 3 different types of OCT scans.
Retina (9mm x 9mm, fovea-centered), ONH (6mm x 6mm, ONH-centered) and raster (14mm x 9mm, including both macular and ONH regions) scans (Fig.1) were acquired from a new multi-mode imaging device that combines ultra-widefield imaging and swept-source OCT (Silverstone, Optos Plc, Scotland). In total, 19 retina/raster and 31 ONH/raster scan pairs were available from a population of healthy and diseased eyes (AMD, DR, macula hole). A deep learning model was used for layer segmentation. Total retina thickness (TRT), nerve fiber layer (NFL) and ganglion cell complex (GCC) macular values, four quadrants of peripapillary NFL thickness and three ONH parameters (disc, cup and neuroretinal rim areas) were automatically derived and analyzed. Pearson’s correlation coefficients were calculated for corresponding pairs of measures. Bonferroni correction for multiple testing was applied to assess statistical significance.
Significant correlation between all 10 investigated measures was observed (Fig.2). Values were especially high for macular thickness (all three above0.95). The same was also true for TRT in all 9 ETDRS subfields (i.e. 984, 0.983, 0.989, 0.988, 0.987, 0.885, 0.988, 0.955, 0.976).
Our results suggest measurements of the same retinal structures are consistent between scan patterns. This is important clinically as it shows patients can be followed accurately if scan patterns are changed over time. They also suggest that a single raster can be used in place of separately captured retina and ONH scans.
This is a 2020 Imaging in the Eye Conference abstract.
Examples of OCT scans: retina (left), ONH (middle), raster (right).
Parameters as mean (standard deviation), root mean square errors (RMSE) and Pearson’s r correlation coefficients between pairs of scans.
This PDF is available to Subscribers Only