Tag Stripes tracking from cardiac MRI by bayesian theory

Abstract

Tag tracking is a pre-step to heart motion reconstruction. In this paper, we present a new tag tracking method based on Bayesian statistical approach, our method works on the basis of tracking with the active grid model, it builds the Markov Random Field (MRF) model according to the prediction of the position of the grid node, and classifies the nodes into two categories considering whether they are in the left ventricle using the EM algorithm, then, different prior distribution and likelihood function are designed for different sorts. The iterated conditional modes (ICM) are utilized to maximize the posterior estimate. The method was validated on several sequences of cardiac systole MRI images. Experiment shows that the method can accurately track the SPAMM tag lines without manually outlining the myocardium, and the grid model can keep its topology during tracking process. © Springer-Verlag 2004.