Purchase this article with an account.
J.J. Jorzik, A. Deckert, S. Schmitz–Valckenberg, A. Bindewald, U. Mansmann, F.G. Holz, FAM Study Group; Computer–assisted Automated Analysis of Digital cSLO Fundus Autofluorescence Images in Advanced Atrophic AMD . Invest. Ophthalmol. Vis. Sci. 2004;45(13):2965.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose:Fundus autofluorescence (FAF) imaging using confocal scanning laser ophthalmoscopes (cSLO) has been shown to be superior to fundus photography for delineating areas of geographic atrophy (GA) of the retinal pigment epithelium. Computer–aided tools can be helpful for monitoring growth of GA over time. Based on our previous work we have assembled and evaluated an advanced customized software for automated detection and quantification of GA areas. Methods:FAF images in vivo were obtained using a cSLO (HRA: exc.488 nm, em.>500 nm). The size of GA areas was assessed in 24 right eyes by two independent readers using novel customized image analysis software including an algorithm for automated identification of interfering vascular structures. Agreement between observers and comparison with the results of previously used software were analyzed by the Bland–Altman design. Results:Retinal vessels block background FAF and, thus, exhibit similar grey levels like the atrophic patches. This normally impedes partially automated segmentation with conventional region finding algorithms. Moreover, many images are illuminated inhomogeneously what complicates segmentation as well. The utilized algorithm reliably identifies vascular structures that interfere with the GA. To detect and disobey discommoding vessels the developed software allows for individual setting of the parameters vessel diameter, length and cross–linking (complexity). Furthermore, a contrast enhancing tool, a sensitive threshold adjustment and relocatable convex and radial hulls facilitate fine–tuning of the actual segmentation of GA areas. Preliminary results suggest that the new method achieves even higher concordance between 2 independent examiners (inter–rater–variability) compared to the previously used software. Difficulties arose from insufficient image quality, retinal vessels touching the atrophic patches tangentially or choroidal vessels with autofluorescent properties underneath the GA area. Conclusions:FAF cSLO imaging allows precise delineation of GA areas. The computer–assisted automated image analysis using new customized software is not only able to reliably filter out vascular structures, but especially enables highly accurate detection and quantification of GA areas. This method will be useful for monitoring effects of future therapeutic interventions.
This PDF is available to Subscribers Only