This paper presents an algorithm for segmenting and measuring retinal vessels, by growing a "Ribbon of Twins" active contour model, which uses two pairs of contours to capture each vessel edge, while maintaining width consistency. The algorithm is initialized using a generalized morphological order filter to identify approximate vessels centerlines. Once the vessel segments are identified the network topology is determined using an implicit neural cost function to resolve junction configurations. The algorithm is robust, and can accurately locate vessel edges under difficult conditions, including noisy blurred edges, closely parallel vessels, light reflex phenomena, and very fine vessels. It yields precise vessel width measurements, with subpixel average width errors. We compare the algorithm with several benchmarks from the literature, demonstrating higher segmentation sensitivity and more accurate width measurement.

This paper concerns the validation of automatic retinal image analysis (ARIA) algorithms. For reasons of space and consistency, we concentrate on the validation of algorithms processing color fundus camera images, currently the largest section of the ARIA literature. We sketch the context (imaging instruments and target tasks) of ARIA validation, summarizing the main image analysis and validation techniques. We then present a list of recommendations focusing on the creation of large repositories of test data created by international consortia, easily accessible via moderated Web sites, including multicenter annotations by multiple experts, specific to clinical tasks, and capable of running submitted software automatically on the data stored, with clear and widely agreed-on performance criteria, to provide a fair comparison.

Performance, simplicity, and robustness of the proposed method demonstrates its suitability for diabetic retinopathy screening applications.

This paper describes REVIEW, a new retinal vessel reference dataset. This dataset includes 16 images with 193 vessel segments, demonstrating a variety of pathologies and vessel types. The vessel edges are marked by three observers using a special drawing tool. The paper also describes the algorithm used to process these segments to produce vessel profiles, against which vessel width measurement algorithms can be assessed. Recommendations are given for use of the dataset in performance assessment. REVIEW can be downloaded from http://ReviewDB.lincoln.ac.uk.

Abstract-Diabetic retinopathy (DR) has been widely studied and characterized. However, until now, it is unclear how different features, extracted from the retinal vasculature, can be associated with the progression of diabetes and therefore become biomarkers of DR. In this study, a comprehensive analysis is presented, in which four groups were created, using eighty fundus images from twenty patients, who have progressed to DR and they had no history of any other diseases (e.g. hypertension or glaucoma). The significance of the following features was evaluated: widths, angles, branching coefficient (BC), angle-to-BC ratio, standard deviations, means and medians of widths and angles, fractal dimension (FD), lacunarity and FD-to-lacunarity ratio, using a mixed model analysis of variance (ANOVA) design. All the features were measured from the same junctions of each patient, using an automated tool. The discriminative power of these features was evaluated, using decision trees and random forests classifiers. Cross validation and out-of-bag error were used to evaluate the classifiers' performance, calculating the area under the ROC curve (AUC) and the classification error. Widths, FD and FDto-Lacunarity ratio were found to differ significantly. Random forests had a superior performance of 0.768 and 0.737 in the AUC for the two cases of classification, namely three-years-pre-