Registration of Diffusion Tensor Images Based on Hybrid Optimization Method

Registration of diffusion tensor images has attracted more and more attention in recent years. In this paper a novel affine registration algorithm for diffusion tensor (DT) MR images that enables explicit analytic optimization of tensor reorientation is presented. The objective function captures both the image similarity and tensor reorientation, which is necessary for warping DT images. The final optimization method seeks a hybrid optimization method. Results show that this method is a stable one, which not only avoids local extrema successfully, but also ensures that the DT orientations remain consistent with the anatomy after image transformations and also improves the average of overlap of eigenvalue-eigenvector.