The Analytic and 12–Month Biological Variability of Retinal Oximetry Measurements in Patients With Non–Neovascular Age Related Macular Degeneration

Abstract

Abstract: : Purpose: To quantify the analytical and biological variability of a new retinal oximeter in patients with age related macular degeneration (AMD). Methods: Retinal oximetry images were obtained at baseline (n=14) and 12 months (n=9) in patients with non–neovascular AMD. Images contained 586 nm (isosbestic) and 605 nm (± 2.5nm) light acquired instantaneously and quantified with a 16–bit scientific digital camera. System noise was quantified with a black image, and subtracted from the retinal image. Optical density ratios (ODRs) were calculated by manually placing a fixed–sized sample window on the edge of arteries and veins. Software automatically sampled an identical window located in the tissue at a fixed distance from the vessel sample. Analytical coefficient of variation (CV) was calculated by analyzing a single baseline image five times. Biological variation was determined from ODRs measured at baseline and 12 months. Results: The analytical CV was 4.02% in retinal arteries, and 1.93% in retinal veins. Removal of 5 outliers reduced the analytic CV to 0.26% and 0.57%, respectively. Twelve–month biological CV was 2.43% in retinal arteries and 0.71% in retinal veins. Conclusions: Outliers in the analysis process may increase variation beyond the level of biological variation. On average, analysis contributes approximately 4% and 2% variability to oximetry measurements in retinal arteries and veins respectively, in patients with non–neovascular AMD. Elimination of camera noise and improved analysis techniques have decreased measurement variability from previously reported levels of 13% and 15% obtained in healthy subjects. Retinal oximetry was found to be a reliable method when used in patients with AMD.

Keywords: imaging/image analysis: clinical • metabolism • image processing