Spline-Based 3D-to-2D Image Registration for Image-Guided Surgery and 3D Electron Microscopy
S. Jonić, P. Thévenaz, C.O.S. Sorzano, M. Unser
Biophotonics for Life Sciences and Medicine, M. Faupel, P. Smigielski, A. Brandenburg, J. Fontaine, Eds., Fontis Media, Lausanne VD, Swiss Confederation, pp. 255–273, 2006.
This paper focuses on techniques for rigid-body registration of a Three-Dimensional (3D) model to a set of Two-Dimensional (2D) x-ray images in Computer-Assisted Orthopedic Surgery (CAOS) and in 3D Electron Microscopy (3D EM). In the first part of the paper, we present a survey of these techniques. In the second part, we describe our strategy for solution of these two problems. Our algorithm relies on a space-based iterative minimization of the least-squares dissimilarity between the x-ray images and their simulations. To allow for a well-defined gradient of the dissimilarity measure when simulating the x-ray images, we use cubic B-splines for interpolation of the volume. We improve the robustness of the algorithm thanks to the multiresolution data pyramids that we compute using cubic B-splines too. In the third part, we show the performance of our algorithm using phantom CAOS and EM data with the ground-truth registration that is known a priori.
@INCOLLECTION(http://bigwww.epfl.ch/publications/jonic0601.html,
AUTHOR="Joni{\'{c}}, S. and Th{\'{e}}venaz, P. and S{\'{a}}nchez
Sorzano, C.{\'{O}}. and Unser, M.",
TITLE="Spline-Based {3D-to-2D} Image Registration for Image-Guided
Surgery and {3D} Electron Microscopy",
BOOKTITLE="Biophotonics for Life Sciences and Medicine",
PUBLISHER="Fontis Media",
YEAR="2006",
editor="Faupel, M. and Smigielski, P. and Brandenburg, A. and Fontaine,
J.",
volume="",
series="",
type="",
chapter="",
pages="255--273",
address="Lausanne VD, Swiss Confederation",
edition="",
month="",
note="")