Morphometric Measurements of The Retinal Vasculature in Fundus Images with Vampire

This chapter reports recent algorithms developed by the vasculature assessment and measurement platform for images of the retina (VAMPIRE) group for vasculature detection and quantification, including recent developments on landmark detection. It focuses on accuracy and validation issues, and, importantly, the conditions for comparing meaningful results from different algorithms. This work is a part of VAMPIRE, which is an international collaboration growing a software suite for automatic morphometric measurements of the retinal vasculature. The chapter focuses on three important retinal image processing (RIA) topics: vessel width estimation, artery-vein classification, and validation. Any quantitative description of the retinal vasculature requires the location of the vascular network. This image segmentation task yields typically a binary vessel map, in which pixels are classified as vessel or not vessel. Validation requires three main components: standardization of validation methodology (protocols), design of public data sets, and standardization of validation metrics