Meeting Banner
Abstract #0266

Exploiting Spatiotemporal Correlations for Accelerating Dynamic 2D Spectroscopic Imaging of Hyperpolarized Pyruvate in the Heart

MAGNA25Kilian Weiss1, Andreas Sigfridsson1, Georgios Batsios1, Marcin Krajewski1, Michael Batel2, Sebastian Kozerke1

1Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland; 2Physical Chemistry, ETH Zurich, Zurich, Switzerland

The investigation of real-time metabolic processes using hyperpolarized compounds requires efficient sampling schedules for sufficient spatiotemporal resolutions. Dynamic hyperpolarized 13C signals in the heart are inherently sparse and their dynamics may be represented by few temporal basis functions. To this end, spatiotemporal acceleration techniques are particularly suited to speed up acquisitions. In this work k-t PCA was implemented and tested based on simulated and hyperpolarized in-vivo data of the rat heart. Accurate reconstructions from up to 8-fold k-t undersampled data have been achieved demonstrating the potential of k-t undersampling approaches for dynamic hyperpolarized imaging of the heart.

Abstract PDF Presentation Video

Keywords

accelerating acceleration acceptable accuracy acid acquisition acquisitions address agreement aims allow apparent appropriate axis back background band bicarbonate biomedical birdcage blood built cardiac chemistry coil combination compressed compressibility concept considered contaminations correlations coverage curves custom decimated dependency derived determined domain dynamic dynamics encoding engineering error errors excitation exploit exploiting extensions field find fold full fully good gradients healthy heart horizontal ideally implemented in vivo incorporate indicated induction inherent inherently injected injection institute investigate lactate leads make material matrix metabolic metabolism mixture model modeled myocardium need noise numerical optimal original parallel partial particular pass pattern peak perfusion physical pixel pool prescribed press previous prior profiles properties proposed proton rapid readout receive reconstructed reconstructing reconstruction reconstructions required resolution respectively root sampled scheme selective sensing series short simulate simulation simulations space sparsity spatial spectral spectroscopic speed square started studies subsequently sufficient suitable suited suppressed surface system tail temporal tool training transmit vain ventricle vertical view width