Feature-based deformable registration of brain images

This paper proposes a novel deformable method of feature-based image registration about two or more different modality images gotten from the same subject, which can automatically determine corresponding points and corresponding segmentation. In this way, we can avoid errors caused by manually or semi-automatically setting corresponding points and improve the exactness and efficiency at the same time. In addition, we can avoid being trapped in local minimization since transformation bases on corresponding segmentation.