Figure 3. Morphological analysis of flat-mounted retina labeled with FITC-conjugated lectin from Bandeira simplicifolia.

(a) Illustration of the procedure followed with “ImageJ” software. (a₁) Selection of seven areas per group, in mid (700x500 px) and extreme (400x300 px) peripheries on mosaics (3007x2904 px) made with microscopic images at low magnification (x4). (a₂) Application of steerable filters for ridge detection in the selected image, with the plug-in “Feature Detector” [57]. (a₃) Transformation of the filtered image into a binary image allowing selection of the retinal vessels to calculate the vascularized area (area covered by vessels out of total area, Av/At). (a₄) Derivation of the vascular skeleton from the binary image, using the procedure “skeletonizes”. This second binary picture was used to calculate the tortuosity index corresponding to the ratio: real vessel length (L2, in red) out of the length of an imaginary straight line on the measured vessel (L1, in blue) as shown in d₁ and d₂. (b–c) Quantitative analysis of the retinal vasculature from control (CTL) and PGF ET-treated eyes. (b) Vascularized area calculated from the binary images which are illustrated in c₁ and c₂. All data: CTL, 18.8%±0.5%; PGF-ET, 23.7%±1.1%; ** p = 0.0034. Mid periphery: CTL, 19.4%±0.7%; PGF-ET, treated eyes 25.1%±0.8%; Bonferroni, ** p<0.01. Extreme periphery: CTL, 18.2%±0.7%; PGF-ET treated eyes 21.2%±1.7%; Bonferroni, ns. (c) Tortuosity index of capillaries calculated from the second binary images. CTL, controls 1.053±0.005; PGF-ET, treated eyes 1.180±0.014; *** p<0.0001; PGF ET/CTL = 1.12.