Non-parametric and non-rigid registration method applied to myocardial gated SPECT

We propose a non-parametric and non-rigid registration method totally unrestricted regarding the allowed deformations. It requires a segmentation contour of the organ of interest in both the image to register and the image template. We assume that the contour template can be deduced from the contour in the image to register by a series of locally normal elementary deformations. This assumption allows us to compute unique deformation vectors. We extend these elementary deformation computations to the whole image domain using level sets (sets of embedded parallel contours with the organ contour as a reference). This series of elementary deformations provides a way to perform an iterative registration by indicating the successive motions from a point in the image to register to its final position in the registered image. Myocardial gated SPECT acquisitions are more susceptible to noise since the average count number per projection is N times lower than equivalent non-gated acquisitions (N is the number of sequence frames). We have applied our method to perform a motion compensation of the frames, i.e. to non-rigidly register N-1 frames with respect to the N/sup th/ one. The N registered frames were then added together resulting in an image with similar noise characteristics to an equivalent non-gated image but without motion-induced blurring. Since our method is based on contour deformation, it guarantees that the registered contour exactly matches the contour template. In this application of sum after motion compensation, we are sure that no point inside the organ in an image will be added to a point of the background of another one. Results on a 3D simulated sequence are presented.