Model Based Detection of Tubular Structures in 3D Images
Abstract
Detection of tubular structures in 3D images is an important issue for vascular medical imaging. We present in this report an extension of a previous work done by the same authors [KMA98]. The following improvements have been done: a better state of the art, a modification of the response function by introducing a coefficient theta whose value is deduced from the model, an experimental study on synthetic images with junctions, tangent structures, structures with varying diameters, and tubular structures with Gaussian-like cross-sections or ``convolved bar-like'' cross-sections, an experimental study on a phantom image and on real X-ray and MRA images.