Évaluation d'une technique de compensation du mouvement des artères coronaires à partir d'une séquence angiographique

Abstract

A method for cardiorespiratory motion compensation of the coronary arteries from a biplane or monoplane angiographic sequence is proposed. The motion is estimated from a 3D/2D registration. The 3D model of the coronary arteries is extracted by triangulation from two angiographic views before the minimally invasive surgery. The cameras parameters are first calibrated by a nonlinear optimization where the reprojection error is minimized. The selection of the arteries is performed manually and the correspondence is calculated from the intersection of the vessel centerline and the epipolar line on the associated view. The cardiorespiratory motion is compensated using a particle filter and the 3D model. Each particle of the system contains an affine deformation used for a 2D projection. This projection is then evaluated with the distance from the artery’s centerline observed in the angiographies. This distance is obtained by a combination of filters on the medical images. A vascularity filter is applied first, followed by a denoising and binarization filter. Finally, the modified version of the distance transform is calculated. An algorithm is also presented to segment the bifurcations of the coronary arteries while minimizing noise. The particles are sampled favoring those minimizing the distance between the arteries and the bifurcation. The particles filter iteratively generate, evaluate and sample the particles with a larger constraint on the deformation parameters. A realistic cardiorespiratory motion simulator is used for two different purposes. It is first used to construct a template deformation parameters for limiting the generation of particles in the system space throughout the cardiac cycle. With known 3D positions of the arteries, the simulator is also used to measure and validate the effectiveness of the suggested cardiorespiratory movement compensation method.

Date	14 Mar 2016
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Luc Duong (Supervisor) & Christian Desrosiers (Co-supervisor)