Multi-modal diffeomorphic demons registration based on point-wise mutual information

Huanxiang Lu, Mauricio Reyes, Amira Serijovic, Stefan Weber, Yasuo Sakurai, Hitoshi Yamagata, Philippe C. Cattin

Abstract: In this paper we propose a variational approach for multimodal image registration based on the diffeomorphic demons algorithm. Diffeomorphic demons has proven to be a robust and efficient way for intensity-based image registration. However, the main drawback is that it cannot deal with multiple modalities. We propose to replace the standard demons similarity metric (image intensity differences) by point-wise mutual information (PMI) in the energy function. By comparing the accuracy between our PMI based diffeomorphic demons and the B-Spline based free-form deformation approach (FFD) on simulated deformations, we show the proposed algorithm performs significantly better.

External IDs:dblp:conf/isbi/LuRSWSYC10