DiVesselSim: CT-Angiography Simulator of Diseased Blood Vessel

Abstract

Accurate determination of vessel parameters is a prerequisite for automated visualization and quantification of healthy and diseased blood vessels. This requires an exhaustive evaluation of different segmentation techniques as a previous step to be used in clinical environment. Normally a large number of test images and manual segmentations used as “gold standard” are required, a synthetic data with segmentation and geometry known would be useful for testing purpose. This work describes a design and implementation of a parametric system that generates synthetic data corresponding to diseased vessel structures as taken from CT-Angiography (CTA). The data consists on a 3D volume made by a set of 2D images (CT slices). The model is based on the properties and physic features of vessels structures (healthy and non healthy) such as cylindrical shape on 3D with elliptical or circular crosssections on 2D, blood vessel and soft tissue density, center of the blood vessel and radio. The generated synthetic data include two different models of diseased blood vessels visually similar to CTA slices from peripheral arteries.