Fast Non-rigid Multimodal Image Registration Using Local Frequency Maps B.C. Vemuri and J. Liu Department of Computer and Information Science and Engineering University of Florida vemuri|jliu@cise.ufl.edu Abstract. In this paper, we present a novel and computationally effi- cient multimodal image registration scheme that is capable of handling non-rigidly misaligned multi-modal data sets. The algorithm is based on a level-set based image registration algorithm applied to local frequency representations of the input images. The algorithm has been tested on several data sets and we present one of these examples. 1 Introduction In this paper, we present a novel method for non-rigidly registering multi-modal images. Our method involves first deriving a brightness “invariant” representa- tion of the image and then using the level-set curve evolution technique for image registration introduced in Vemuri et.al., [1] on these “invariant” representations of the input images. 2 Local Frequency Representation & Matching The Gabor Filter is a well-known quadrature filter. It achieves the theoretical lower bound of the uncertainty principle. Local frequency computation can be achieved by Gabor filtering the input data and then computing the gradient of local phase followed by summing the squared magnitude of local frequency maps over a discrete set of orientations for a fixed frequency [2]. We present and example local frequency representation in figure 2 which depicts a pair of T1 & T2 MR slices and their corresponding local frequency maps. Given two local frequency maps F 1 (X ) and F 2 (X ), we want to find the non- rigid transformation between them. This can be achieved by decomposing the non-rigid deformation into incremental rigid transformation and using a moving grid implementation of the following evolution equation: V =[F 2 (X ) - F 1 (V (X )] ∇F 1 (V (X )) ‖∇F 1 (V (X ))‖ with V (X, 0) = 0 (1) where V =(u, v, w) T is the displacement vector at X and the operation V (X )= (x - u, y - v,z - w). For an efficient implementation of this equation, see [1]. W. Niessen and M. Viergever (Eds.): MICCAI 2001, LNCS 2208, pp. 1363–1365, 2001. c  Springer-Verlag Berlin Heidelberg 2001