Assessment of retinal vascular geometry in normal and diabetic subjects

Abstract

Diabetic retinopathy is the most common microvascular complication of diabetes mellitus and the leading cause of blindness in persons from age 20 to 74. The relative risk of blindness in persons with diabetes has been reported to be 5.2 times the risk of those without diabetes. The fundus abnormalities described in diabetic retinopathy result from structural damage to the microvasculature wall with subsequent leakage or as a result of retinal ischaemia with secondary overproduction of vascular growth factors. Several clinical and screening classifications schemes have been developed to categorize and quantify the severity of each of the retinopathic features based on the degree of retina involvement. The ultimate goals of these classification schemes have been to provide a system by which the natural history of the disease and the risk of progression of retinopathy and visual loss can be identified and the subsequent response to interventions can be evaluated to improve patient care. The present strategies for dealing with diabetic retinopathy address retinopathy that is already established. However, recent studies - supported by computer based imaging analysis – have focused on changes in retinal vascular caliber and demonstrated various associations with increased risk of diabetes and predicted the onset of microvascular retinal complications. This suggests that other structural and geometrical parameters might also be utilised, which can provide more information regarding the retinal vascular network. Few studies have reported different changes in retinal vascular geometry with age, systemic hypertension and peripheral vascular diseases. The objective of this thesis is to analyse the retinal vascular geometrical features in normal subjects and evaluate its role in diabetic subjects with different stages of diabetic retinopathy. For this purpose, a semi-manual vascular analysis technique is designed to measure and analyse the different retinal vascular geometrical parameters and ratios. The developed technique performance and precision is compared to other available manual and semi-manual vascular analysis techniques.The various sources of variability in retinal geometrical measurements are then evaluated, including observer’s measurement errors, variations in image capture, and potential short term changes in the subjects’ vascular geometrical features. The second step of this work is to perform a detailed analysis of the retinal vascular geometry in normal subjects, including the topographic distribution of different geometrical measurements across the fundus, the effect of different demographic and clinical factors, and the stability of measurements between both eyes. The next step evaluates the retinal vascular geometry in diabetic subjects with different grades of diabetic retinopathy to determine any changes of geometrical features with advancement of retinopathic stages. The results demonstrate significant associations of changes in vascular structural and geometrical features with increased stages of diabetic retinopathy. Finally, the predictive value of retinal vascular geometry analysis and its practical role on the individual level is analysed for a sample of subjects who progressed from no retinopathy to proliferative retinopathy as compared to a sample of subjects with no sign of progression. The preliminary results suggest that geometrical changes trend can be detected on the individual level with progression of diabetic retinopathy and those differences can be noted between progressors and non-progressors at baseline. In conclusion, this thesis describes novel retinal vascular geometrical markers indicative of establishment of advancing diabetic retinopathy, together with a potential predictive role in determining risk of future progression to proliferative retinopathy.