Bayesian Multimodality Non-rigid Image Registration via Conditional Density Estimation

Abstract

We present a Bayesian multimodality non-rigid image registration method. Since the likelihood is unknown in the general multimodality setting, we use a density estimator as a drop in replacement to the true likelihood. The prior is a standard small deformation penalty on the displacement field. Since mutual information-based methods are in widespread use for multimodality registration, we attempt to relate the Bayesian approach to mutual information-based approaches. To this end, we derive a new criterion which when satisfied, guarantees that the displacement field which minimizes the Bayesian maximum a posteriori (MAP) objective also maximizes the true mutual information (with a small deformation penalty) as the number of pixels tends to infinity. The criterion imposes an upper bound on the number of configurations of the displacement field. Finally, we compare the results of the Bayesian approach with mutual information, joint entropy and joint probability approaches on synthetic data and simulated T1 and T2 2D MR images.