Abstract
Purpose: :
Fundus autofluorescence (FAF) images can be used for semi-automated detection and measurements of areas of geographic atrophy (GA) due to age-related macular degeneration (AMD). The aim of this study was to determine intra- and interobserver reproducibility for the quantification of areas of GA in a reading center setting.
Methods: :
Standardized FAF images were recorded with a confocal scanning laser ophthalmoscope (Spectralis/Heidelberg Retina Angiograph, Heidelberg Engineering, Germany). Two independent readers quantified total size of atrophy in 59 eyes with GA using semi-automated image analysis software. A total of 28 were analyzed a second time by each observer one month later in order to assess intraobserver variability. Minimum size for atrophic patches was set at 0.05 mm2. Corresponding infrared reflectance images were viewed side by side to FAF images for confirmation of GA lesion boundaries and foveal involvement. Agreement was assessed by Bland-Altman test.
Results: :
The average GA area was 7.21 mm2. The first quartile was 2.56 mm2, the median 6.19 mm2 and the third quartile 10.69 mm2. The mean differences of intraobserver agreements were -0.06 mm2 (95% confidence interval [CI], -0.19 to 0.07) for reader 1 and -0.04 mm2 (95% CI, -0.15 to 0.08) for reader 2. Interobserver agreement showed a mean difference of 0.08 mm2 (95% CI, -0.07 to 0.22). Reasons for disagreement included difficulties in assessing foveal involvement, presence of confluent peripapillary atrophy and small GA satellites.
Conclusions: :
Standardized semi-automated analysis of atrophy quantification using FAF images represents an accurate method with a high degree of reproducibility. Accurate measurements of GA areas plays a critical role for determining progression over time and, thus, for interventional clinical trials with therapies aiming at slowing enlargement of GA over time.
Keywords: age-related macular degeneration • imaging/image analysis: clinical • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology