Purchase this article with an account.
Michael J Allingham, Daniel Izatt, Qing Nie, Eleonora M Lad, Priyatham S Mettu, Scott W Cousins, Sina Farsiu; Robust, easy-to-use, semiautomated software quantifies lesion rim area hyperautofluorescence and predicts progression of geographic atrophy.. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):2829.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Several methods of fundus autofluorescence (FAF) image analysis have been used to predict progression of geographic atrophy (GA), however these methods are time intensive and require highly trained image graders. Additionally, controversy remains as to whether increased autofluorescence (AF) associated with GA is correlated with lesion progression. We wished to develop a rapid, easy-to-use, quantitative means of image analysis to segment GA and associated hyper-AF to examine the hypothesis that increased AF at lesion borders predicts progression of GA.
MATLAB based semiautomatic software was developed to segment and quantify GA lesion size, rate of progression and rim area focal hyper-AF (RAFH). A retrospective analysis of serial FAF images taken from 49 eyes of 30 subjects with GA was performed. Correlation between RAFH and progression of GA was analyzed using Spearman correlation. Comparison between RAFH tertiles was performed using generalized estimating equations and Kruskal-Wallis testing. To assess reproducibility, the calculated growth rate and RAFH were compared between two retina clinician graders.
Average GA lesion size was 7.55 mm2 and the average rate of progression was 0.16 mm2/month. RAFH was positively correlated with rate of GA progression with a Spearman correlation coefficient of 0.49 (p<0.001) for grader 1 and 0.54 (p<0.001) for grader 2. Subjects in the middle or highest tertile of RAFH were at greater risk of progression (p=0.005 for medium vs. low and p=0.001 for high vs. low). Median difference in RAFH between graders was 1.1% (IQR 0.02- 1.3%). Median difference in calculated lesion growth rate was 2.8% (IQR 0.87-8.8%). Importantly, stratification of subjects into RAFH tertiles was 97% identical between graders.
Progression of GA is positively correlated with RAFH. Our image analysis software was easy to-use as a high school student learned to use it effectively in a few hours. This software permits accurate, reproducible quantification of GA lesions and stratification of risk of progression based on RAFH. Our findings suggest that hyper-AF at GA lesion borders is predictive of lesion progression. RAFH may serve as an imaging biomarker for progressive GA and could be used to identify patients at increased risk of rapid progression for inclusion in future trials or for therapeutic intervention.
This PDF is available to Subscribers Only