A New Approach of Image Registration for Biomedical Images Using Saliency Information

In this paper we propose a Markov Random Field based Automatic Registration method. This is an elastic registration method that uses the combination of saliency and gradient information. This Intensity-based registration of images is done by linear transformations, based on a discrete Markov random field (MRF) formulation. Here, the challenge arises from the fact that optimizing the energy associated with this problem requires a high-order MRF model. Currently, methods for optimizing such high-order models are less general, easier to use, and efficient, than methods for the popular second-order models. Automatic registration of dynamic contrast enhanced magnetic resonance (DCE-MR) images is a challenging task due to the rapid changes of the images which are characterized by intensity changes over time, thus posing challenges for conventional intensity-based registration methods. Saliency information contributes to a contrast invariant metric to identify similar regions inspiteof contrast enhancement. Saliency is used in optimization framework to identify relevant pixels for registration, thus reducing the computation time. Experimental results on real patient images demonstrate superior registration accuracy with a combination of saliency and gradient information over other similarity metrics.