Fundus image based blood flow simulation of the retinal arteries

Abstract

Computational fluid dynamic (CFD) simulations can help to understand the hemodynamics of the retinal vascular network and the microcirculation. Systemic diseases, like hypertension and diabetes, change the geometry of the vasculature in the retina and these changes can be seen with fundus photography. Furthermore, these changes are indicators of cardiovascular diseases. The aim of this study is to create a plane 2D model of the retinal arterial network based on a high-resolution fundus photograph and to perform a CFD simulation. The blood vessels were segmented from the image with the Frangi filter method. A structural fractal tree was implemented to calculate the outflow boundary conditions representing the peripheral vascular bed. With the Frangi filter method and the high-resolution fundus image a comprehensive model of the visible retinal artery network could be achieved. The simulation results show realistic velocity and pressure distributions of the retinal blood flow in a healthy retina compared to in-vivo measurements in the literature. This work is an initial step towards creating comprehensive patient-specific models of the retinal vasculature based on readily available fundus photography.