Purchase this article with an account.
Valentina Franco-Cardenas, Christopher J. Gee J. Gee, David Appap, Kentaro Nishida, Adriana Ramirez, Adriana Perez-Rovira, Kris Zutis, Steven D. Schwartz, Emmanuel Trucco, Jean-Pierre Hubschman; Assessment of Retinal Non-Perfusion in Vascular Retinal Diseases using Single versus Summarized Ultra Wide-Field Fluorescein Angiography Images. Invest. Ophthalmol. Vis. Sci. 2012;53(14):2074.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Image summarization via RERBEE (robust efficient registration via bifurcations and elongated elements) is a novel feature-based registration algorithm able to correct local deformations in high-resolution ultra wide-field fluorescein angiogram (UWFFA) sequences of the retina. An overlay of a UWFFA sequence is used to produce a single summarized image. Creating summarized images may be used in the future for automated retinal image analysis (ARIA) of ischemic index (ISI), instead of analyzing each image of the UWFFA. Calculating an ISI, using a single UWFFA image, is useful for assessing neovascular complications and macular edema in retinal vascular pathologies; however, it is unknown if a summarized image provides equivalent results.
Two sets of images (single vs. summarized) were provided to 6 graders. Each set consisted of 5 branch retinal vein occlusion (BRVO), 5 central retinal vein occlusion (CRVO) and 5 diabetic retinopathy (DR) images. Each summarized image was obtained via RERBEE overlay of 5 images of the different phases of the UWFFA sequence. Graders were asked to circle the non-perfused areas. A Wilcoxon signed rank test was done to determine significant difference between grading single versus summarized images by disease. Intraclass correlation coefficient (ICC) was used to determine inter-grader agreement.
The detected area of non-perfusion is grater when using single images vs summarized images in BRVO, CRVO and DR. The retinal non perfusion areas measured with a single vs a summarized image are equivalent in BRVO and CRVO (p=0.12, p=0.67), but nor for DR (p=0.005). Overall there is a better agreement when using the single image.
Calculating an ISI in summarized images obtained via RERBEE seems to be a viable option. In the future, we consider that using summarized images for ARIA of non-perfusion may be considered in retinal vein occlusions but not for DR.
This PDF is available to Subscribers Only