Structural image representation for image registration

We propose a structural image representation and show its relevance for multi-modal image registration. Structural representation means that only the structures in the image matter and not the intensity values of their depiction. The representation is formulated as a dense descriptor. We specify three properties an optimal descriptor for structural registration has to fulfill: locality preservation, structural equivalence, and discrimination. The proposed entropy images are an approximation to such a representation. We improve their discriminative potential by integrating spatial information in the density estimation. We evaluate entropy images for rigid, deformable, and groupwise multi-modal image registration and achieve very good results in terms of both speed and accuracy. Finally, entropy images seamlessly integrate into existing registration frameworks and allow an efficient registration optimization.