Deformable Registration of MR Brains with Best Features

Introduction Deformable registration is very important formedical image analysis and so far various methodshave been proposed in decades1-4]. HAMMER reg-istration algorithm5]defines an attribute vector in-cluding intensity, edge type, and geometric mo-ment invariants as a signature of each point, to re-duce ambiguity in correspondence matching duringthe image registration. Another characteristic ofHAMMER is the hierarchical matching mecha-nism, which helps avoid the warping being trapped…

Deformable Registration of MR Brains with Best Features

A learning-based deformable registration method was presented for MR brain images. First, best geometric features are selected for each location and each resolution, in order to reduce ambiguity in matching during image registration. The best features are obtained by solving an energy minimization problem, which requires the features to be distinctive around the neighboring points and consistency across training samples. Secondly, the set of active points is hierarchically selected based on their saliency and consistency measurements during registration, which helps to produce accurate registration results. Finally, by incorporating those learned results into the framework of HAMMER, great improvement in both real data and simulated data is achieved.