Summarising the retinal vascular calibres in healthy, diabetic and diabetic retinopathy eyes

Abstract

Retinal vessel calibre has been found to be an important biomarker of several retinal diseases, including diabetic retinopathy (DR). Quantifying the retinal vessel calibres is an important step for estimating the central retinal artery and vein equivalents. In this study, an alternative method to the already established branching coefficient(BC) is proposed for summarising the vessel calibres in retinal junctions. This new method combines the mean diameter ratio with an alternative to Murray's cube law exponent, derived by the fractal dimension,experimentally, and the branch exponent of cerebral vessels, as has been suggested in previous studies with blood flow modelling. For the above calculations, retinal images from healthy, diabetic and DR subjects were used. In addition, the above method was compared with the BC and was also applied to the evaluation of arteriovenous ratio as a biomarker of progression from diabetes to DR in four consecutive years, i.e. three/two/one years before the onset of DR and the first year of DR. Moreover, the retinal arteries and veins around the optic nerve head were also evaluated. The new approach quantifies the vessels more accurately. The decrease in terms of the mean absolute percentage error was between 0.24% and 0.49%, extending at the same time the quantification beyond healthy subjects.

Evaluation of geometric features as biomarkers of diabetic retinopathy for characterizing the retinal vascular changes during the progression of diabetes

“…No significant results were observed for any features, when we compared either the pair first-year-pre-DR/firstyear-of-DR, which comes almost in line with the early results presented in previous study [6], or the pair of threeyears-pre-DR/two-years-pre-DR.…”

“…These changes are believed to start affecting the vascular geometry early, before the first stages of DR. Until now no comprehensive effort has been made, in order to study the effects that diabetes has to the geometry of the retina in patients that progressed to diabetic retinopathy, in multiple time intervals of diabetes. Moreover, the investigated features were limited, for specific periods of time, or using different population as control group that does not provide us with indications on how the retinal vasculature changes within each patient during diabetes [4][5][6].…”

Exploiting the Retinal Vascular Geometry in Identifying the Progression to Diabetic Retinopathy Using Penalized Logistic Regression and Random Forests

“…A few follow-up studies have been conducted, studying similar periods of diabetes, without though including in any classification system or evaluating features like central retinal vein/artery equivalent or tortuosity, which is the purpose of this study [18,4,20,19,21].…”

A new unified framework for the early detection of the progression to diabetic retinopathy from fundus images

“…The first one is the widely used method as introduced in [38], and the second one is a more recent and more accurate approach, proposed in [37]. The area of interest, in which all measurements are taken, includes the region where the edges of the vessels course through at 0.5 to 1.0 disc diameters from the OD margin.…”