Meeting Banner
Abstract #2019

MRS Sparse-FFT: Reducing Acquisition Time and Artifacts for in vivo 2D Correlation Spectroscopy

Lixin Shi1, Ovidiu C. Andronesi2, Haitham Hassanieh1, Badih Ghazi1, Dina Katabi1, Elfar Adalsteinsson1, 3

1Electrical Engineering and Computer Science Department, Massachusetts Institute of Technology, Cambridge, MA, United States; 2Martinos Center for Biomedical Imaging, Radiology Department, Massachusetts General Hospital, Harvard Medical School, Boston, MA, United States; 3Harvard-MIT Division of Health Sciences and Technology, Cambridge, MA, United States

In-vivo localized 2D Correlation Spectroscopy (COSY) enables detection of new molecular biomarkers of disease. The challenges for this method are, however, long acquisition times and truncation artifacts along the f1 frequency dimension that may obscure the cross-diagonal peaks. We propose to address these challenges by adapting the recently developed sparse-FFT algorithms, which optimize reconstruction of sparse signals in the Fourier domain. Our preliminary results demonstrate that sparse-FFT can reduce measurement time of 2D COSY by almost a factor of three, eliminates t1 truncation artifacts resulting from the ringing tails of the diagonal peaks, and improves SNR and resolution of cross-peaks.

Abstract PDF Poster PDF

Keywords

acquisition adapting allow allows almost applications artifacts assignment audience benefit berger biased biomedical body brain characterize clinical comparing components compressed computer consecutive correlation cross crowded customized dealing described designed despite detect developed developers development diagonal dimension disease disentangle division domain electrical eliminate encoding engineering enhancing evolution exploit fall fast filtering finding fractional frequencies frequency full functions general great grid half health hence improve in vivo increments indicate instead introduce labeled largely laser lends leverage limitation linear localized mandatory medical medicine molecular naturally needed nelson optimized overlap peaks phantom portion positions potential prediction preliminary price propose radiology random reconstruct reconstruction reduce reducing related remove requires resolution respectively ringing robust routine samples sampling scanner school science sciences scientists selectively sensing several soda sparse sparsity spectra spectral spectroscopy spectrum spread stemming subtract suffers summarized superior suppression tail tails takes taking target technology terms towards transform translational trio truncation typically unambiguous useful utility validation volunteers whole windowing