Motion Analysis of the Right Ventricle From MRI Images Edith Haber 1, Dimitris N. Metaxas 2, and Leon Axel 3 1 Department of Bioengineering 2 VAST Lab, Department of Computer and Information Science 3 Department of Radiology University of Pennsylvania, Philadelphia, PA 19104-6389, USA edith@seas, upenn, edu http://~, cis.upenn, edu/~edith/ Abstract. Both normal and abnormal right ventricular (RV) wall mo- tion is not well understood. In this paper, we use data from tagged MRI images to perform the first 3D motion study of the entire right ven- tricle to date. Our technique is an adaptation of a physics-based de- formable modeling methodology that was successfully used on the left ventricle(LV). As opposed to the previous approach, currently we use segmented contours to generate the geometry, 1D tags for our input data (due to the thinner RV), and localized degrees of freedom (DOFs) with finite elements. Although we build a biventricular model, our results focus on method validation and visualizing clinically useful parameters that describe RV wall motion. 1 Introduction Abnormal motion of the RV serves as an indicator of several types of heart disease, such as RV ischemia and hypertrophy [3]. However, there is no in-depth knowledge of the RV motion and its correlation to the various diseases. One reason that researchers do not agree about the exact pumping mechanism of the normal RV may be regional variations in contraction patterns occuring at different phases of systole. Since studies have found that RV contraction varies with increases in pressure and volume, and ischemia [3], a method for accurately assessing RV wall motion can both answer questions into its normal function and can be used as an indicator of various diseases. The RV receives blood from the right atrium and pumps it into the pul- monary artery. It appears crescent-like in a cross-sectional view and, unlike the LV, is difficult to define with any parameterized 3D shape. The RV shares a septum with the LV, while its outer free wall is in mechanical contact with the pericardium and the lungs. The RV cavity can be conceptually separated into an inflow tract, a highly-trabeculated apical portion, and a relatively smooth out- flow tract [6] (Fig. 1). The 3ram-thick free wall is thin relative to the 7ram-thick LV free wall. This relative thinness and complex geometry make it difficult to capture RV wall motion using regular imaging modalities. This paper presents a new methodology for modeling and analyzing the RV shape and motion from MRI-tagged data. We make significant modifications to