[Static vessel analysis in nonmydriatic and mydriatic images]

Static vessel analysis is a method to determine the diameter of retinal vessels in images of the ocular fundus. The suitability of non-mydriatic and mydriatic images for that method and the influence of mydriasis on the results were examined.

In the prospective study, 30 eyes of 15 patients (10 women, mean age 51.6 +/- 13.2 years) were examined. At first, 3 images were taken of each eye with the retinal camera Topcon NW 200 (magnification 1). After pupil dilation with tropicamid eye drops, 3 more images were taken using the Topcon and 3 others using the system Visualis (IMEDOS, Jena/Germany, FF450plus, 535-561 nm, 30 degrees image, 1840 x 1360 pixel). The vessel diameters were measured with the software Vesselmap2 (IMEDOS). The investigator assigned vessels to arteries or veins and their diameters were calculated automatically by the software. There is the possibility to define the vessel edge manually in cases of a poor image quality. The calculation of the central retinal arterial and venous equivalent (CRAE, CRVE) as well as the arterio-venous ratio (AVR) were made according to the formula of Parr-Hubbard. Furthermore, the nasal retinal vessels > 60 microm were examined to estimate the influence of tropicamid on the vessel diameter.

Because of insufficient illumination and poor contrast in 21 % of the non-mydriatic images not all vessels could be detected automatically. Additionally, fewer vessels could be detected in 7 % of the non-mydriatic images compared to the mydriatic images. The average coefficient of variation of CRAE and AVR of each triplet of images was higher in non-mydriatic images (2.6 % and 3.2 %, respectively) than in mydriatic images of the Topcon (1.8 %; 2.3 %) and the FF 450 (1.7 %; 1.8 % ANOVA p < 0.05). No significant differences were found between the various examination methods for both the coefficient of variation of CRVE (1.9 %; 1.8 %; 1.7 %) and the average values of CRAE, CRVE and AVR. With regard to their diameters, the nasal retinal arteries and veins > 60 microm, were depicted sufficiently in all images, and only differed insignificantly between the three methods.

The quality of non-mydriatic images is often lower than that of mydriatic images. This fact can account for the high variance of measured parameters in the non-mydriatic images. The depiction of all relevant vessel segments is a precondition for the image-based analysis. An influence of the mydriasis caused by tropicamid on the retinal vessel diameters > 60 microm was not found.