A Multicomponent Approach to Nonrigid Registration of Diffusion Tensor Images

摘要

We propose a nonrigid registration approach for diffusion tensor images usinga multicomponent information-theoretic measure. Explicit orientationoptimization is enabled by incorporating tensor reorientation, which isnecessary for wrapping diffusion tensor images. Experimental results ondiffusion tensor images indicate the feasibility of the proposed approach and amuch better performance compared to the affine registration method based onmutual information in terms of registration accuracy in the presence ofgeometric distortion.