A Cost Efficient Approach for Automatic Non-Rigid Registration of Medical Images
Sami Dhahbi, Walid Barhoumi, Ezzeddine Zagrouba · 2009
Equipe de Recherche en Syste`mes Intelligents en Imagerie et Vision Artificielle (SIIVA-ISI)Institut Supe´rieur d’Informatique, 2 Rue Abou Rayhane Bayrouni, 2080 Ariana, Tunisia{sami.dhahbi,walid.barhoumi}@[email protected]://www.isi.rnu.tnAbstract. A common approach for non-rigid medical image registrationis thehierarchical image subdivision-based strategy. In this approach, images are pro-gressively subdivided, locally registered, and elastically interpolated. Althoughthis approach seems to be among the fastest approaches for non-rigid registra-tion, computation time is still a real challenge. This work deals with this prob-lem and proposes a new hierarchical strategy. To reduce computational com-plexity, we propose to combine in the same framework the hierarchical imagesubdivision-based strategy with a Gaussian pyramid. The hierarchical subdivi-sion method ensures that the registration process deals with small and large de-formations, whereas the use of Gaussian pyramid decreases the computation timeenormously. The proposed framework is preliminary validated in the context ofmonomodal registration by matching breast mammograms and MRI brain imageswith simulated deformations. Registration quality is evaluated by using imagedifferences, mean square error, peak signal to noise ratio and correlation coef-ficient. Complexity study and experimental results show tha t the proposed ap-proach reduces considerably the computation cost meanwhile maintaining com-parable accuracy.