Meeting Banner
Abstract #0709

Dramatic Speedup in 1D-, 2D- & 3D-MRS Scan Times with Linear Algebraic Modeling (SLAM)

MAGNA25Yi Zhang1, 2, Refaat E. Gabr1, Michael Schr1, 3, He Zhu1, 4, Peter Barker1, 4, Robert G. Weiss1, 5, Paul A. Bottomley1, 2

1Division of MR Research, Department of Radiology, Johns Hopkins Univesity, Baltimore, MD, United States; 2Department of Electrical and Computer Engineering, Johns Hopkins Univesity, Baltimore, MD, United States; 3Philips Healthcare, Cleveland, OH, United States; 4F. M. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, MD, United States; 5Division of Cardiology, Department of Medicine, Johns Hopkins Univesity, Baltimore, MD, United States

Scan-time is a central concern for chemical shift imaging (CSI). While model-based MRS reconstruction methods could reduce scan times significantly, their in vivo application is limited and focused on suppressing inter-compartmental leakage. Here, spectroscopy localization with linear algebraic modeling (SLAM) is introduced to dramatically speed-up scan-time. SLAM uses a minimal number of phase-encoding steps that are selected from central k-space, to reconstruct average spectra from pre-selected sample compartments. We demonstrate that SLAM yields essentially the same spectra as compartmentally averaged 1D, 2D and 3D CSI spectra, but 4-, 16- and 100-fold faster, respectively. The signal-to-noise ratio cost was ≤50%.

Abstract PDF Presentation Video

Keywords

accuracy achieve achieved acquisitions advantage advantages algebraic application applied apply approximates auxiliary background barker best blue brain cardiac cardiology cast central chemical chest closely comp compartment compartmental compartments computer containing cost datasets define dimension dimensional disk distinguishable division done dong double dramatic electrical eliminate elimination encodes encoding engineering entire equation essentially experiment faster focused fold fraction functional grant grid hardly heart highest human humans identical implemented in vivo institute inter johns just knowledge known lateral leakage least limited linear lipid loan localized long major math matrix medicine modeling noise offering offset operator operators overlay peter phantom phantoms phosphate plus practice priori problem process product propose radiology realized reconstruct reconstructed reconstruction reduce reduced reduction reductions remains respectively retains rows scalp scout segment selected slam slim sloop solved space spatial spectra spectroscopy spectrum speed speedup steps suffice support suppressed suppression system theory transform triple truncating unique unknown unknowns vector ventricle volume water whether whole yields