Background: This study compared the sensitivity and specificity of stereoscopic digital photography of the retina through a dilated pupil with a 45° nonmydriatic camera and Joint Photographic Experts Group (JPEG) compression of the images with the sensitivity and specificity of 35-mm slide film photography in the identification of age-related macular degeneration (AMD). Methods: Consecutive patients with a diagnosis of AMD were enrolled. Stereoscopic retinal images of the disc,macula and temporal macula were captured with a digital 45° nonmydriatic camera (then compressed into JPEG format) and with a standard fundus camera and slide film. A single retinal specialist graded both image formats in masked fashion, at least 1 month apart, using a modified Age-Related Eye Disease Study (AREDS) severity scale.The digital images were displayed on a monitor and viewed with the use of liquid crystal display shutter glasses and stereo imaging software.The film images were mounted on a light box and graded with the use of a stereoviewer. Primary outcome measures included the presence or absence of AMD pathological features. Positive and negative predictive values (PPVs and NPVs), sensitivity, specificity and weighted κ w statistics were calculated. Results: We photographed 203 eyes (of 103 patients) with both digital and slide film cameras. Correlation of the 2 image formats was substantial in identifying AREDS level 3a or greater (κ w = 0.64, standard error = 0.08, PPV = 0.95, NPV = 0.66, sensitivity = 0.93, specificity = 0.74) and excellent in identifying level 4b or greater (κ w = 0.83, standard error = 0.05, PPV = 0.81, NPV = 0.98, sensitivity = 0.94, specificity = 0.94). Interpretation: High-resolution stereoscopic, mydriatic, 45° digital images captured with a nonmydriatic camera and JPEG compressed correlate well with stereoscopic slide film photographs in the identification of moderate to advanced AMD (AREDS level 3a or greater).
Read full abstract