Fundus digital image processingautomated segmentation of the main retinal anatomical structures

Abstract

Diabetic retinopathy (DR) is a chronic disease, caused by complications of diabetes mellitus, which nowadays constitutes the primary cause of blindness in people of working age in the developed world. In spite of DR is not curable, its progression can be controlled if it is detected in the early stages. However, early detection of DR is a difficult task since affected patients do not perceive symptoms until visual loss develops, which occurs in the later disease stages. This thesis is framed within a research project which aim is the development of a comprehensive system for the automated diagnosis of DR by means of digital retinal image analysis. One of the main tasks in the implementation of such a system is the automated segmentation of the main anatomical structures of the retina. This is the main goal of this thesis, the development of new high accurate, robust and fast methodologies for the optic disc, vascular tree and macula segmentation. Moreover, in the context of the automated segmentation of the vascular tree, a new function for global quality evaluation of these segmentations is also proposed. The presented segmentation methodologies enhance the accuracy shown by the most outstanding techniques present in the literature. Therefore, this work constitutes a step forward in the development of automated systems for the early detection of DR.