Registration and matching of projection imaging and tomographic imaging. Application to X-rays angiographies and magneti
In this paper, we propose a new method for matching vascular imaging modalities without using external frame or external landmarks. First, 3D reconstruction of part of the cerebral vascular tree is performed using Magnetic Resonance Angiography (MRA). The
- PDF / 579,283 Bytes
- 6 Pages / 481.89 x 685.984 pts Page_size
- 35 Downloads / 165 Views
CARS/Springer. All rights reserved.
419
Registration and matching of projection imaging and tomographic imaging. Application to X-rays angiographies and magnetic resonance angiographies. M. Vermandel a, b, C. Kulik b, lY. Gauvrita, C. Vasseurb , J. Rousseaua aInstistut de technologie Medicale, Pav. Vancostenobel, CHRU de Lille, 2 avo O. Lambret, 59037, Lille cedex. b I 3 D, bat. P2, USTL, 59 655, Villeneuve 0' Ascq, France
Abstract In this paper, we propose a new method for matching vascular imaging modalities without using external frame or external landmarks. First, 3D reconstruction of part of the cerebral vascular tree is performed using Magnetic Resonance Angiography (MRA). Then, this volume is projected on Digital Subtracted Angiography (DSA) image until its best position and orientation are found. Results show satisfactory robustness and accuracy. Keywords: registration, matching, angiography.
1. Introduction This study presents a new approach for registration and matching of projection and tomographic imaging. For illustrating the applied method, this paper deals with X-Rays and Magnetic Resonance vascular cerebral images. Several authors [I, 2] describe methods on 2D and 3D registration. Those methods can be classified as two different approaches which are intensity-based registration[ I] and feature-based registration[2]. The method described in this paper is based on a 3D modeling of the volume of interest and re-projection algorithm using conic projection properties. Developments presented have been applied on both phantom and patient images. Preoperative MRA scans have been used to generate a 3D model of the vascular tree. DSA images were acquired in standard pre-operative conditions.
2. 2D/3D Registration procedure 2.1 Principle Matching images needs several preliminary steps as registration or researching landmarks changing relations. In order to register MRA and DSA, we propose a method illustrated below (Fig. 1).
CARS 2002 - H. U. Lemke, M. W. Vannier; K. Inamura, A.C. Farman, K. Doi & J.H.c. Reiber (Editors)
'CARS/Springer. All rights reserved.
420 MRJ\ Dala
V"lulU-= R"""'>Il'i.lnk:ll~lft
USA
~~:I'J).:nlll ll(.ln
0.."
Fig. 1: Block diagram illustrating the methodology applied.
First, MRA and DSA images of considered volume are acquired. Then, a 3D model of the initial volume is reconstructed from MRA. This reconstructed model becomes the anatomic reference linking the two modalities. Then, the model is digitally projected applying conic projection properties. The resulting image is compared to a segmented DSA image through a similarity measure. Finally, the optimal position and orientation of the model in the DSA space is iteratively found. At the end of the process and considering that the structure configuration is known in the MRA referential (thanks to the reconstruction step), the complete matching between the two modalities is immediately obtained. 2.2 Volume reconstruction The model is reconstructed from the MRA dataset. We first generate a Maximum of Intensity Projection (MI
Data Loading...
