Publication: Retinal vessel density measurement as a tool for microvascular research
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Subject LCSH
Retina--Diseases
Subject ICSI
Call Number
Abstract
This study aimed to evaluate the vessel density parameter as another measurement to study the morphological changes of retinal vessel. This study assessed the reliability of the vessel density parameter; provided the empirical evidence of fractal dimension (DF) as an indirect measure of the vessel density; investigated the effect of aging on the parameters and conducted a preliminary analysis of the ability of the parameters to distinguish between normal and diseased retinal states. A total of 200 coloured right eye fundus images were analyzed to obtain two quantitative parameters which were vessel density and DF. The coloured fundus images were transformed into black and white using MyVessel software. A circular area of approximately 2.6 optic disc radii surrounding the center of the optic disc was cropped using GNU Image Manipulation Program (GIMP) version 2.6.12 and the non-vessel fragments were deleted using the same software. The values of retinal vessel density and DF were obtained from the FracLac software version 2.5. The findings showed that the intragrader reliability of vessel density and DF as assessed using intraclass correlation coefficient (ICC) were 0.9975 (95% CI = 0.9954 - 0.9987) and 0.9645 (95% CI = 0.9340 - 0.9811), respectively. For intergrader reliability, the ICC were 0.9910 (95% CI = 0.9831 - 0.9952) and 0.9725 (95% CI = 0.9484 - 0.9854) for vessel density and DF, respectively. A strong correlation between vessel density and DF parameters was detected with Pearson’s correlation coefficient, R of 0.942 (p<0.001). In response to increase in age, both vessel density and DF parameters showed declining trend, suggesting rarefaction of the retinal vessel. Preliminary analysis on the sensitivity of vessel density and DF parameters to differentiate between normal and diseased retinal states was obtained from the area under the receiver operating characteristic curve (AUC). The AUC of vessel density and DF were 0.91 (p<0.001) and 0.89 (p<0.001), respectively. This study demonstrated the ability of the vessel density parameter to describe the difference between younger and older age groups. The vessel density parameter also showed high association with currently used geometrical parameter of retinal vessel, DF. Therefore, vessel density parameter may be considered as another parameter for quantitative analysis of retinal vessel.