Accurate high-speed spatial normalization using an octree method

The goal of regional spatial normalization is to remove anatomical differen ces between individual three-dimensional (3-D) brain images by warping them to match features of a standard brain atlas. Fall-resolution volumetric sp atial normalization methods use a high-degree-of-freedom coordinate transfo rm, called a deformation field, for this task. Processing to fit features a t the limiting resolution of a 3-D MR image volume is computationally inten sive, limiting broad use of full-resolution regional spatial normalization. A highly efficient method, designed using an octree decomposition and anal ysis scheme, is presented to resolve the speed problem while targeting accu racy comparable to current volumetric methods. Translation and scaling capa bilities of oct;ree spatial normalization (OSN) were tested using computer models of solid objects (cubes and spheres). Boundary mismatch between tran sformed and target objects was zero for cubes and less than 1% for spheres. Regional independenee of warping was tested using brain models-consisting of a homogenous brain volume with one internal homogenous region (lateral v entricle). Boundary mismatch improved with successively smaller octant-leve l processing and approached levels of less than 1% for the brain and 5% for the lateral ventricle. Five 3-D MR brain images were transformed to a targ et 3-D brain image to assess boundary matching Residual boundary mismatch w as approximately 4% for the brain and 8% for the lateral ventricle, not as good;as with homogeneous brain models, but similar to other results. Total processing time for OSN with a 256(3) brain image (1-mm voxel spacing) was less than 10 min. (C) 1999 Academic Press.