Key points are not available for this paper at this time.
We previously developed a realistic phantom for the cardiac motion for use in medical imaging research. The phantom was based upon a gated magnetic resonance imaging (MRI) cardiac study and using 4D non-uniform rational b-splines (NURBS). Using the gated MRI study as the basis for the cardiac model had its limitations. From the MRI images, the change in the size and geometry of the heart structures could be obtained, but without markers to track the movement of points on or within the myocardium, no explicit time correspondence could be established for the structures. Also, only the inner and outer surfaces of the myocardium could be modeled. We enhance this phantom of the beating heart using 4D tagged MRI data. We utilize NURBS surfaces to analyze the full 3D motion of the heart from the tagged data. From this analysis, time-dependent 3D NURBS surfaces were created for the right (RV) and left ventricles (LV). Models for the atria were developed separately since the tagged data only covered the ventricles. A 4D NURBS surface was fit to the 3D surfaces of the heart creating time-continuous 4D NURBS models. Multiple 4D surfaces were created for the left ventricle (LV) spanning its entire volume. The multiple surfaces for the LV were spline-interpolated about an additional dimension, thickness, creating a 4D NURBS solid model for the LV with the ability to represent the motion of any point within the volume of the LV myocardium at any time during the cardiac cycle. Our analysis of the tagged data was found to produce accurate models for the RV and LV at each time frame. In a comparison with segmented structures from the tagged dataset, LV and RV surface predictions were found to vary by a maximum of 1.5 mm's and 3.4 mm's respectively. The errors can be attributed to the tag spacing in the data (7.97 mm's). The new cardiac model was incorporated into the 4D NURBS-based Cardiac-Torso (NCAT) phantom widely used in imaging research. With its enhanced abilities, the model will provide a useful tool in the study of cardiac imaging and the effects of cardiac motion in medical images.
Segars et al. (Thu,) studied this question.