Purchase this article with an account.
P. A. Parsons-Wingerter; Fractal Analysis of Retinal Arterial Tree in Diabetic Retinopathy. Invest. Ophthalmol. Vis. Sci. 2010;51(13):4696.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To examine changes in the fractal dimension (Df) of the human retinal arterial tree with progression of diabetic retinopathy (DR) from mild nonproliferative DR (NPDR) to proliferative DR (PDR). Fractal analysis is used to quantify the space-filling capacity of branching patterns such as the retinal vasculature. Previous studies showed a decrease in Df of mild NPDR compared to the normal human retina.
For prospective evaluation of retinal vascular patterns in patients with NPDR, binary (black/white) arterial trees were isolated from 50° digital fluorescein angiograms by semi-automatic computer processing. Df was calculated for binary and skeletonized images by box-counting at a power of two using the VESsel GENeration Analysis (VESGEN) software.
In arterial trees, Df of the binary and skeletonized images of retinal arteries increased from 1.574 and 1.344 for mild NPDR (n = 3) to 1.601 and 1.376 for PDR (n = 2). The p-values from a one-tailed Student’s t-test of equal variance for expected differences (α = 0.05) were 0.037 and < 0.01 for the binary and skeletonized images. As in previous studies, Df of skeletonized images was a sensitive indicator of change in the space-filling capacity of vascular branching patterns.
Fractal analysis of retinal arterial trees revealed that Df increased with progression of DR from mild NPDR to PDR. These changes were statistically more significant in the skeletonized images than in the binary images.
This PDF is available to Subscribers Only