Clustering and detecting 2D motion fields in ultrasonic images based on regression with respiration and MRFs

In consideration of organ motion, we propose a new method by which to discriminate motion patterns and to detect the 2D motion field simultaneously using ultrasonography and respiratory signals. In the proposed method, we use discrete MRFs to represent localization of same motion regions. Similarity of motion is defined as equality of the regression coefficients between the motion and the respiration. A computationally stable algorithm is constructed using the EM algorithm strategy, and the effectiveness of the method is confirmed through real image experiments.