Variational Approach to Cardiac Motion Estimation for Small Animals in Tagged Magnetic Resonance Imaging

Abstract

Monitoring cardiac motion in the stage of small animal study is very important in cardiac research. This paper presents a variational approach to estimating the heart motion of small animals imaged by magnetic resonance (MR) tagging. Small animals have much faster heart beats than human, so their cardiac sequences are temporally undersampled, leading to the aperture problem when reconstructing the cardiac motions. To overcome this difficulty, we adopt the prior knowledge of motions on the myocardial boundaries that were determined in the preprocessing. In addition, we utilize the high gradients of intensities on the tag lines to derive the motions through the cardiac cycle. We formulate the problem in the framework of energy minimization. Variational calculus gives us the Euler-Lagrange equations to seek the minimum. The results produced by our approach are better than the existing optical flow based method [1] and the harmonic phase method [2]. The evaluation suggests that our approach will be more suitable for the small animal studies.