A novel approach for global registration of medical images based on learning the prior appearance model

A new approach to align an image of a medical object with a given prototype (reference object) is proposed. Visual appearance of the images, after equalizing their signals, is modeled with a new variant rotation and scaling Markov-Gibbs random field with pairwise interaction model. Similarity to the prototype (reference object) is measured by a Gibbs energy of signal co-occurrences in a characteristic subset of pixel pairs derived automatically from the prototype (reference object) using our previous Linear Combination of Discrete Gaussians (LCDG) probabilistic model. An object is aligned by an affine transformation maximizing the similarity by using an automatic initialization followed by gradient search. Experiments confirm that our approach aligns complex objects better than popular conventional algorithms.