Aorta centerline smoothing and registration using variational models

Abstract

In this work we present an application of variational techniques to the smoothing and registration of aorta centerlines. We assume that a 3D segmentation of the aorta lumen and an initial estimation of the aorta centerline are available. The centerline smoothing technique aims to maximize the distance of the centerline to the boundary of the aorta lumen segmentation but keeping the curve smooth. The proposed registration technique computes a rigid transformation by minimizing the squared Euclidean distance between the points of the curves, using landmarks and taking into account that the curves can be of different lengths. We present a variety of experiments on synthetic and real scenarios in order to show the performance of the methods.