Purchase this article with an account.
L.B. Wilson, B.D. Madjarov, C.C. Liu, G.–S. Ying, A.L. Ells, A. Fielder, K. Cocker, M. Mills, S. Davidson, G.E. Quinn; Can Digital Rop Images Be Optimized for the Remote Reader? . Invest. Ophthalmol. Vis. Sci. 2006;47(13):5298.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To determine if optimal levels of brightness and contrast of digital ROP images exist that can be standardized for presentation to the remote reader.
Using proprietary software to adjust 50 digital images (25 posterior pole and 25 peripheral) taken from babies at ROP risk, 6 readers were required to manipulate image brightness and then contrast to acheieve optimal image quality. Readers were then required to state whether the optimized image was sufficient for grading. The inter–grader agreement in the brightness and contrast was assessed by intra–class correlation (ICC) coefficient and its 95% confidence interval.
After optimization, the proportion of images that the 6 readers judged sufficient for grading ranged from 48% to 78%. 84% of the time at least 4 of the 6 readers agreed that the final optimized image was sufficient for grading. The graders agreed moderately well on the brightness optimization. ICC (95% C.I) for brightness was 0.65 (0.56–0.74) for all 50 images, 0.64 (0.52–0.78) for posterior pole images, and 0.65 (0.52–0.79) for peripheral retinal images. The graders agreed poorly on the contrast optimization. ICC was 0.05 (0.01–0.15) for all 50 images, 0.11 (0.02, 0.27) for posterior pole images, and 0.007 (–0.07, 0.15) for peripheral retinal images.
Readers tended to select similar levels of brightness for an image, regardless of whether posterior or peripheral retina. However the level of brightness selected varied from image to image. Selection of contrast level for each image tended to vary widely across both readers and images. This suggests that image quality varied for digital images, perhaps due to the difficulty of obtaining images or variability of images from the premature baby’s eye. At present, standardization of image brightness and contrast for presentation to the remote reader does not seem feasible.
This PDF is available to Subscribers Only