Aggregated Motion Estimation for Parallel Real-Time MRI

Many modern applications of magnetic resonance imaging (MRI) require high temporal and spatial resolution. Because the data acquisition speed is fundamentally limited by physical and physiological constraints it is important to find approaches to reduce the amount of acquisitions without deteriorating the image quality. Many existing solution strategies are based on strong a priori assumptions about the unknown object, which often introduces undesirable bias in the image. To overcome this difficulty, we have recently introduced a novel algorithm that estimates the deformation between nearby frames and incorporates this information into the reconstruction process. Our method is not restricted to affine or rigid motions, and does not need additional measurements. In the present study, we present further reconstruction results from phantom and in vivo cardiac measurements demonstrating the increased temporal and spatial resolution of our approach compared to state-of-the art algorithms.