Comparison of nonmydriatic digitized video fundus images with standard 35-mm slides to screen for and identify specific lesions of age-related macular degeneration.

PURPOSE

The purpose of this study was to compare nonmydriatic digitized images obtained using a digital imaging system (resolution of 640 x 480 pixels) with 35-mm slide images for detecting specific findings of age-related macular degeneration (AMD) and to evaluate its usefulness as a screening tool in detecting signs of AMD.

METHODS

Seventeen consecutive patients (33 eyes) underwent digital color imaging (with a nonmydriatic, 45-degree, fundus camera attached to a digital back) and standard 35-mm, 30-degree retinal color photography of the fundus: posterior pole, nasal retina, and temporal retina. The images were later reviewed for the presence or absence of specific retinal findings. The images were not compressed. Primary outcome measures included the presence or absence of drusen, hard exudate, choroidal neovascularization (CNV), subretinal hemorrhage, retinal pigment epithelial (RPE) changes, subretinal fibrosis, pigment epithelial detachment (PED), and subretinal fluid. Presence of drusen, with or without any one of the other findings, and presence of disciform scar or geographic atrophy were positive indications in screening for AMD.

RESULTS

Agreement between image type was highest for PED (97%), CNV (91%), and subretinal fibrosis (91%); and lowest for RPE changes (63%). Sensitivity, specificity, positive predictive value, and negative predictive values were determined using the 35 mm-slide images as the reference for comparison. Sensitivity ranged from 40% (hard exudates) to 75% (subretinal hemorrhage). Specificity ranged from 88% (drusen) to 100% (hard exudate, CNV, RPE changes, PED). Positive predictive value ranged from 67% (subretinal hemorrhage) to 100% (hard exudate, CNV, RPE changes, PED). Negative predictive value ranged from 44% (drusen) to 97% (PED). For the purposes of screening for any evidence of AMD, the system was 70% sensitive.

CONCLUSION

This digital fundus imaging system with 640 x 480 pixel resolution has low sensitivity and high specificity, as compared with 35-mm slide images, for detection of early AMD, but higher sensitivity for late findings (CNV, scar, atrophy) of AMD. Because of sensitivity for detecting any AMD coupled with the low sensitivity for detecting CNV, the system is not useful for evaluating AMD patients who require close follow-up and who are at risk for more severe visual loss.