Markov Random Field model based multimodal medical image registration

A new method based on Markov Random Field (MRF) model to register multimodal medical image is proposed. First, a multimodality intensity transformation or mapping function, which is estimated from the marginal peaks in a joint histogram of two images, is introduced. The transformation function is applied to one image to create a virtual image that hat has similar intensity correspondence characteristics to the other one, of a different modality. Then, using the original two image matrices and the transferred two image matrices, we formulate a new MRF energy function comprising a data term which is similar to a distance function and a smoothness term that penalizes local deviations. In optimization step, a quasi-Newton optimization algorithm is used to find the minimal value of the MRF energy function. The test results show that the proposed algorithm has better performance in both accuracy and robustness to noise, on a series of 2D MRI and CT images.