Article
Evaluation of a semi-automated image software for quantification and progression analysis of geographic atrophy
Search Medline for
Authors
Published: | June 15, 2011 |
---|
Outline
Text
Background: A novel semi-automated software enables quantification and progression analysis of age related macular degeneration (AMD)-associated geographic atrophy (GA) based on fundus autofluorescence (FAF) imaging. We systematically evaluated software performance and reader training .
Methods: Three-field FAF, near-infrared (NIR) and blue reflectance mode (BR) images were recorded (Heidelberg Retina Angiograph/Spectralis, Heidelberg Engineering, Germany) at baseline, after 6 and 12 months according to protocol. 30 randomly chosen subjects were analyzed by seven experienced readers in blinded fashion, and two new readers after initial instruction on the software (RegionFinderTM 2.0, Heidelberg Engineering), featuring a database export function, semi-automated detection of GA areas after definition of seed points on central FAF images, shadow/vessel recognition and graphic display on grading reports. We recorded alignment and processing limits as well as inter- and intraobserver variability.
Results: At baseline the mean size of atrophy was 5.96 mm2 (range: 1.80–15.87). The mean progression rate was 1.25 mm2/year (0.42–2.93). Mean difference of interobserver agreement ranged between -0.25 to 0.30 mm2 for baseline visit and between -0.14 to 0.11 mm2/year for atrophy progression. Processing limits occurred due to misalignment in case of low imaging quality. For lesion boundary discrimination NIR and BR were of complementary use for the inexperienced reader, e.g. regarding foveal GA sparing or pronounced pigment attenuation.
Conclusions: Quantification of atrophy and progression using this software is accurate and reproducible, even among first time users. Naturally, imaging quality limits software performance. Short instruction to the software would already render precise and time-efficient GA measurements in clinical context like in context of interventional trials on new therapies for GA deceleration.