Diffusion MRI-based tractography, which is built by connecting the principal components of the estimated water diffusion pattern, is used to elucidate neuronal connectivity.However, the anatomical accuracy of the method is affected by factors such as noise and imaging misalignments.In this manuscript, we present a method to clean diffusion datasets by rotating the diffusion patterns according to a transformation matrix, that is in turn obtained in the fractional anisotropy domain.We demonstrated the accuracy and feasibility of our method by improving brain images of a mouse and also by creating a diffusion template in a neonates framework.