Abstract
Purpose:
To evaluate an automated drusen detection and quantification software tool from RTVue® Spectral Domain Optical Coherence Tomography (SD-OCT) (Optovue, Inc., Fremont, CA), comparing it to traditional manual colored photography drusen quantification, towards development of an automated assessment of the AREDS Severity Scale and Simple Scale.
Methods:
After approval of the study protocol by a Johns Hopkins University School of Medicine Institutional Review Board, 18 eyes of 9 patients of a retina specialist (NMB) with a clinical diagnosis of the intermediate stage of AMD in both eyes, as defined by AREDS, who had received an Optovue 3D OCT and color fundus photographs as part of standard care on the same day were reviewed retrospectively. One eye was excluded due to difficulty in grading fundus photograph. Drusen area calculation within the 6mm ETDRS grid was performed using both an automated algorithm of the OCT and manual grading of large drusen by two independent retina specialists (YW, VC) for each one of the 1, 3, and 6 mm ETDRS circles. The averaged results of the two graders were compared to those of the automated algorithm, and the intergrader variability was evaluated by Bland-Altman statistics.
Results:
The mean (SD) drusen area in 1, 3, and 6 mm ETDRS circles was 0.14 (0.15), 1.10 (0.83), and 1.83 (1.17) mm2, respectively, for the averaged manual grading and 0.13 (0.17), 0.73 (0.67), 1.24 (0.93) mm2 for the automated OCT grading. The drusen area calculated automatically was consistently smaller than the manually calculated corresponding drusen area, with the difference increasing with increasing drusen area (best described as y=-0.2525x-0.2126, where y is the difference between OCT and the average of two graders and x is the mean of OCT and the average of two graders for the 6 mm ETDRS circle). Correlation (r2) recorded between the averaged graders drusen area and the automated drusen area calculations for the 1, 3, and 6 mm ETDRS circles was 0.002, 0.719, and 0.702, respectively.
Conclusions:
The quantification of drusen area using automated RTVue® SD-OCT drusen detection and quantification tool yielded good agreement when compared to traditional manual grading. Refinements of this tool may simplify the process of grading severity of AMD in clinical trials and clinical practice in the future.
Keywords: 412 age-related macular degeneration •
504 drusen •
550 imaging/image analysis: clinical