Artery-Venous Classification

Retinal vessel morphology parameters such as vessel diameter, vessel shape, branching pattern, bifurcation angles and tortuosity can be studied to diagnose the different disease which manifests themselves through changes in the retina. The effect of various diseases is different on veins and arteries, so it is very essential to classify retinal vessels correctly as arteries or veins. For example, ‘plus’ disease in retinopathy of prematurity (ROP) may result in increase in arterial tortuosity relative to that of veins indicating the need for preventative treatment. Arterial narrowing, venous dilatation, and resulting decrease in artery-to-venous width ratio (AVR) may predict the future occurrence of a stroke event or a myocardial infarct. Unfortunately, the detection of minute changes in vessel width or tortuosity specific to arteries or veins may be difficult in a visual evaluation by an ophthalmologist or by a semi-automated method, which is laborious in clinical practice. Therefore, an automated identification and separation of individual vessel trees and the subsequent classification into arteries and veins should be developed for vessel specific morphology analysis.
The successful analysis of retinal vessels depends on the correct detection of the retinal vessels to diagnose and monitor the progress of different diseases. The presence of noise, uneven illumination and contrast challenges in the fundus images hinders the correct Automatic Detection of the vessel network.

The goal of this project is to develop algorithms for the classification of retinal blood vessels into arteries and veins.


  • Thorough review on the research/methodologies done in the Artery-Venous Classification in retinal images.
  • To study the effective morphological features of retinal vessels and use them in machine learning algorithms to do medical screening of diseases.
  • Make use of the acquired knowledge in devising a novel idea in automation of the vessel tree formation to perform arteries and veins classification.
  • Use of standard/experts’ results on the available datasets to compare and validate the results.

Areas of Application

  • Early detection of hypertension retinopathy
  • Detection of artery nicking
  • Identification of venous bleeding

Public datasets

Posted in Medical Image Analysis, Projects.