Problèmes inverses, application à la reconstruction compensée en mouvement en angiographie rotationnelle X

Abstract : This work is an application of the inverse problem theory to the motion compensated 3-D reconstruction of coronary arteries from Rot-X data. We first investigate the inverse problem in finite and infinite dimensions. For the finite-dimensional case, we focus on the tomographic inverse problem modeling, by defining the principles of a voxel basis and of the projection matrix, in order to obtain a matrix inverse problem formulation. We also investigate the notion of dynamic tomography and its related issues. Our discrete formulation, using the voxel basis, allows us to introduce any given diffeormorphism support deformation function in the matrix inverse problem formulation, provided that this deformation is a priori known. Our last chapter demonstrates how to estimate coronary arteries motion from ECG gated Rot-X projections, with a coronary 3-D deformable model. The motion is modeled by a B-spline model for point-matching registration. Once the motion is estimated, the tomographic reconstruction is performed at a reference cardiac state throughout a penalized least-squares optimization process including the motion, the penalty term being defined by favouring high intensity values for voxels in the neighborhood of the 3-D centrelines and low intensity values for all other voxels. This method has been tested on simulated data based on 3-D coronary centrelines previously extracted from a MSCT sequence.
Liste complète des métadonnées

https://tel.archives-ouvertes.fr/tel-00361396
Contributor : Alexandre Bousse <>
Submitted on : Saturday, February 14, 2009 - 12:19:18 PM
Last modification on : Wednesday, May 16, 2018 - 11:23:17 AM
Document(s) archivé(s) le : Tuesday, June 8, 2010 - 7:20:21 PM

Identifiers

  • HAL Id : tel-00361396, version 1

Collections

Citation

Alexandre Bousse. Problèmes inverses, application à la reconstruction compensée en mouvement en angiographie rotationnelle X. Traitement du signal et de l'image. Université Rennes 1, 2008. Français. ⟨tel-00361396⟩

Share

Metrics

Record views

541

Files downloads

369