New computer vision techniques to differentiate arteries and veins in retinal angiography images

This basic research project follows a working liine initiated in previous projects towards the development of an algorithm to automatically compute the arteriovenous index, a diagnostic value proposed by ophthalmologists to use in previous detection and following of illnesses reflected in eye vasculature: hypertension, cardiovascular and cerebrovascular illnesses, diabetes, ... One of the steps to do in this process is to differentiate arteries from veins in the angiography image in order to compute the index automatically. As a result, this project develops the previous steps to the automatic measurement: the optic disk extraction and the segmentation of the vascular tree.

Objectives

  • Optic disk detection, as a previous step to the evolution of a deformable contour inside the eye vasculature.
  • Extraction of the eye vasculature using a deformable model.
  • Exploring techniques to differentiate arteries and veins.