Simultaneous Estimation of Dynamic Cardiac MR Images and Deformation Maps

Purpose: Many spatiotemporal models based on image intensity have been used to accelerate cardiac image acquisition [1,2]. Their performance is limited by the fact that the local pixel intensity can abruptly change due to cardiac motion, and particularly the motion of tissue boundaries. In this work, we consider spatiotemporal models based on 2D deformation and propose an approach for simultaneous estimation of dynamic cardiac MR images and deformation maps from under-sampled k-space data. The working assumption is that myocardium deforms continuously and smoothly, and may result in a lower order model that fully captures the dynamics.