Purchase this article with an account.
Ronald P Danis, Amitha Domalpally, Yijun Huang; Standardized assessment of RPE and retina thickness from SDOCT volume scans in a large multicenter trial. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):5937.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate the methodology and reproducibility for assessment of retinal pigment epithelium thickness (RPE) and central retina thickness in a large clinical trial setting using a custom spectral domain optical coherence tomography (SD-OCT) software application after conversion of files to a standard format.<br />
In a clinical trial of geriatric patients under study for non-ocular disease, a preliminary analysis of the baseline OCT volume scan images from 946 eyes (473 participants) enrolled from 82 reading-center certified clinical sites were centrally collected for analysis. Exclusions included intermediate AMD or worse or other confounding ocular abnormalities and age range was >54 to <86 years . After conversion to DICOM OPT format, macular volume scan files collected from 4 different SDOCT machines were displayed in the same software and segmented for ILM, RPE and Bruch’s membrane layers in a unified display and analysis platform using an algorithm developed at the University of Wisconsin’s Fundus Photograph Reading Center (Huang, et al, PLoS One. 2013 Dec 26;8(12):e82922) . Reading center graders reviewed each volume scan and edited boundary line placement if required. Reproducibility was tested through masked regrading to assess intergrader variabilty.<br />
Of 947 eyes, the proportion of ungradable scans was 5% for center retinal thickness and 2% for center RPE thickness. The standardized assessment of mean center RPE thickness was 33.3 microns (SD 4.6) and mean center retinal thickness was 244.1 microns (SD 27.4). From masked regrading, the Weighted Variance of intergrader difference in center subfield RPE thickness was 0.682 microns and for center subfield retina thickness 38.57 microns. Intraclass correlation coefficients were 0.97 for central subfield RPE thickness and 0.90 for center subfield retina thickness.<br />
Standardized analysis of retinal and RPE thickness from multiple types of SDOCT machines is feasible in large clinical trials via conversion to standard file format, storage in a single database, and standardized assessment using the same segmentation algorithm. With reading center certification of the clinical sites, the ungradable image rate was low in this elderly population. Reproducibility for reading center determination of center subfield RPE and retina thickness was excellent .<br />
This PDF is available to Subscribers Only