Purchase this article with an account.
V. Cosatto, B. Taylor, G. Liew, Y. P. Zhang, W. Hsu, M. L. Lee, P. Mitchell, T. Y. Wong, J. J. Wang; Does the Photographic Angle of Incidence Alter the Measured Fractal Dimension of the Retinal Vasculature?. Invest. Ophthalmol. Vis. Sci. 2008;49(13):2108.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Fractal dimension of the retinal vasculature is a new global measure of its architecture and may reflect systemic vascular health. To validate new software for fractal dimension measurement from retinal photographs, we examined whether subtle differences in the photographic angle of incidence affected fractal dimension measures.
A random sample of 30 pairs of stereoscopic photographs of the right optic disc was selected from Blue Mountains Eye Study participants. These photographs were taken at slightly different angles of incidence to obtain stereoscopic views. The IRIS-Fractal program was used to measure fractal dimension. This program traces the vessel network to produce a skeletal image and then quantifies fractal dimension by a box-counting method. A grader refines the tracings by removing background noise (peripapillary atrophy, choroidal vessels, etc) before computing fractal dimension. We assessed images before and after applying a crop function which restricted the image measured to a defined circular zone 3.5 optic disc radii from the optic disc center. Paired student’s t-test (p <0.05) and Pearson correlation coefficient (R2) were performed using SAS.
Mean fractal dimensions were 1.506 (95% confidence interval, CI, 1.501-1.511) and 1.514 (95% CI 1.508-1.520), respectively for the 1st and 2nd photos of the stereo pairs as measured before image cropping (p=0.014). The corresponding means were 1.467 (95% CI 1.461-1.473) and 1.466 (95% CI 1.461-1.472), respectively, after image cropping (p=0.79). The correlation coefficient (R2) between 1st and 2nd photos of the stereo pairs was 0.36 before, and 0.58 after image cropping.
This PDF is available to Subscribers Only