Department of Mathematics and Statistics,
Arizona State University
Friday,
February 12, 2004, 12:15 p.m. in GWC 110
Jinling Yang
CSE Department, ASU
Biomedical Image Registration using B-spline Pyramids
Abstract
The automatic registration algorithms we discuss use hierarchical
B-splines to minimize the difference between a reference and a source data set.
The minimization is performed either by a variation of the Marquardt-Levenberg
algorithm for nonlinear least-square optimization or by a Principal Axes
Transformation and B-spline control points iterative refinements. The deformation
model is a global three-dimensional affine transformation. We obtain excellent
results for the registration of intramodality magnetic resonance imaging
(MRI) data. We conclude that the multiresolution refinement strategy is faster
and more robust. In addition, several improvements are briefly discussed.
For further information please contact:
mittelmann@asu.edu