Meeting Banner
Abstract #1670

Non-Cartesian Calibrationless Parallel Imaging Reconstruction: Application to Hyperpolarized 13C Imaging

Markus Durst1, 2, Ulrich Koellisch1, Jonathan I. Sperl2, Eliane Weidl3, Florian Wiesinger2, Axel Haase1, Rolf F. Schulte2

1IMETUM, Technische Universitt Mnchen, Munich, Bavaria, Germany; 2GE Global Research, Munich, Bavaria, Germany; 3Department for Nuclear Medicine, Technische Universitt Mnchen, Munich, Bavaria, Germany

Due to the irreversible T1-decay of the magnetization, scan time is crucial for hyperpolarized 13C MRI. In contrast to existing autocalibrated parallel MRI algorithms, calibrationless parallel imaging does not require a fully sampled k-space region as used in existing autocalibrated methods, and therefore yields a further reduction of acquisition time. The basic idea is to reconstruct the data with low-rank matrix completion. This existing approach was succesfully adopted to hyperpolarized 13C spiral imaging. Simulations were performed as well as a reconstruction of in-vivo rat data, results were comparable to existing autocalibrated parallel MRI methods, nevertheless allowing a faster acquisition.

Abstract PDF Poster PDF

Keywords

achieved acquiring acquisition acquisitions additional additionally adequate agents alanine aliasing allowing allows almost application applied array avoiding back biochemical blocks buffalo calibration cellular channel coil combination comparable completely completion concentrations conditions consistent consisting constructed consuming conversion covering crucial dataset decay decomposition dependent depletion designed detecting duration durst enhancement entries examine excitation excitations existing exploits fast find fold frequencies fully funded generated global goal grants great grid hence ideal implemented in vivo insight interleaves interpolation irreversible iterative kept knowledge labeled lactate least like linearly magnetization male manner maps matrix measured metabolic metabolite metabolites minimal missing modulated need noise nominal nuclear offers omitted outer overlapping oversampling parallel performance potential preserved prior problem processes promising property proton pulse rank rapid reconstructed reconstruction reduced reduction reeder removed requiring sampled scanner schemes seems sensitivities served simulated simulating simulation singular slice smooth solved space spectroscopic spiral spirit squares structured successfully theory threshold thresholded trajectories trajectory transition unique uptake whereas