Lecture notes in computer science vol:2717 pages:122-131
Second international workshop on biomedical image registration - WBIR'03 location:Philadelphia, Pennsylvania, USA date:June 23-24, 2003
We propose a multimodal free-form registration algorithm that matches voxel class labels rather than image intensities. Individual voxels axe displaced such as to minimize the Kullback-Leibler distance between the actual and ideal joint probability distribution of voxel class labels, which axe assigned to each image individually by a previous segmentation process. We evaluate the performance of the method for intersubject brain registration with simulated deformations, using a viscous fluid model for regularization. The root mean square difference between recovered and ground truth deformations is smaller than I voxel.
D’Agostino E., Maes F., Vandermeulen D., Suetens P., ''An information theoretic approach for non-rigid image registration using voxel class probabilities'', Lecture notes in computer science, vol. 2717, pp. 122-131, 2003 (Proceedings second international workshop on biomedical image registration - WBIR 2003, June 23-24, 2003, Philadelphia, Pennsylvania, USA).