Cardiac motion simulator for tagged MRI

Abstract

Describes a computational simulator for use in cardiac imaging using tagged magnetic resonance imaging. The simulator incorporates a 13-parameter model of left-ventricular motion due to Arts et al. (1992) and applies it to a confocal prolate spherical shell, resembling the shape of the left ventricle. Using parameters determined in other work, our model can be made to assume a configuration representing one of 60 phases in the cardiac cycle. In this paper we determine the inverse motion map analytically, allowing pointwise correspondences to be made between two points at any two times. Using this mathematical relationship, we simulate the (tagged) magnetic resonance imaging process using a standard (tagged) spin-echo imaging equation. Image sequences can be synthesized at arbitrary orientations at any phase. We currently synthesize a SPAMM tag pattern with arbitrary spatial frequency, but other patterns can be readily incorporated. To accommodate two-dimensional motion estimation algorithms, we have created a two-dimensional simulator which restricts the three-dimensional motion to two dimensions. In either two or three dimensions, a true motion is output so that motion estimation algorithms can be compared against the truth. We conclude with a simple demonstration of the performance of the simulator.