Eui Seok Han1, Ki Ho Park. 1. Department of Ophthalmology, Seoul National University College of Medicine, Seoul, Korea.
Abstract
PURPOSE: To compare three types of digital images for assessment of retinal nerve fiber layer (RNFL) defects: red-free RNFL images, color nonmydriatic images, and digitally converted nonmydriatic red-free fundus images. DESIGN: Diagnostic test comparison. METHODS: Ninety-five image sets of three type images were evaluated. Converted images were acquired from nonmydriatic digital images by converting them into red-free, monochromatic images with graphics software. Taking RNFL images as standards, the sensitivities/specificities of nonmydriatic digital color and converted images were evaluated for detecting wedge-shaped defect (WSD) or diffuse atrophy. RESULTS: In terms of WSD, the sensitivity/specificity of converted images were 85.7%/95.5%, and those of nonmydriatic were 83.9%/91.0%. Regarding diffuse atrophy, the sensitivity/specificity of converted images were 60.0%/97.8%, and those of nonmydriatic were 60.0%/98.9%. Intragrader/intergrader agreements for converted images were 0.849/0.793 (WSD) and 0.417/0.614 (diffuse atrophy). CONCLUSIONS: Converted nonmydriatic fundus images without pupillary dilation are as effective as conventional red-free RNFL images for detecting WSD.
PURPOSE: To compare three types of digital images for assessment of retinal nerve fiber layer (RNFL) defects: red-free RNFL images, color nonmydriatic images, and digitally converted nonmydriatic red-free fundus images. DESIGN: Diagnostic test comparison. METHODS: Ninety-five image sets of three type images were evaluated. Converted images were acquired from nonmydriatic digital images by converting them into red-free, monochromatic images with graphics software. Taking RNFL images as standards, the sensitivities/specificities of nonmydriatic digital color and converted images were evaluated for detecting wedge-shaped defect (WSD) or diffuse atrophy. RESULTS: In terms of WSD, the sensitivity/specificity of converted images were 85.7%/95.5%, and those of nonmydriatic were 83.9%/91.0%. Regarding diffuse atrophy, the sensitivity/specificity of converted images were 60.0%/97.8%, and those of nonmydriatic were 60.0%/98.9%. Intragrader/intergrader agreements for converted images were 0.849/0.793 (WSD) and 0.417/0.614 (diffuse atrophy). CONCLUSIONS: Converted nonmydriatic fundus images without pupillary dilation are as effective as conventional red-free RNFL images for detecting WSD.