Meeting Banner
Abstract #3354

Greater Acceleration Through Sparsity-Promoting GRAPPA Kernel Calibration

Daniel S. Weller1, Jonathan R. Polimeni2, 3, Leo Grady4, Lawrence L. Wald2, 3, Elfar Adalsteinsson1, Vivek Goyal1

1EECS, Massachusetts Institute of Technology, Cambridge, MA, United States; 2A. A. Martinos Center, Dept. of Radiology, Massachusetts General Hospital, Charlestown, MA, United States; 3Dept. of Radiology, Harvard Medical School, Boston, MA, United States; 4Dept. of Image Analytics and Informatics, Siemens Corporate Research, Princeton, NJ, United States

When applying GRAPPA at high accelerations, it is not always feasible to acquire sufficiently many auto-calibration signal (ACS) lines to properly calibrate the interpolation kernels. The proposed calibration method employs regularization promoting joint sparsity of the coil images that would be reconstructed. This method improves reconstruction quality and increases the total acceleration that is achievable with GRAPPA.

Keywords

able accelerated acceleration accelerations achieve acquiring acquisition affected affine aliased aliasing always amplification analytics approximated array blocks broad calibrated calibrating calibration called channel coil consistently constraining convolution covariance cropped curves depict dept develop effective effectively either emulate enabling encoded enough equations even evident excitation explore expressed fast fewer field fits formed frequency full general greater head hospital impose included informatics inset instead institute insufficient interpolated interpolation investigated isotropic joint kernel kernels least like limit linear magnitude maintains manually many matrix measured medical metric minimize minutes mitigate mitigates noise none operations operator outnumber parallel peak perceived prior promoting proposed quality quantitative radiology reasonably receive reconstruct reconstructed reconstruction reduce reduces regularization regularized remains representative requires requiring residual resolution retained samples scanner school sense sensitivities slice source space sparsity spirit squares straightforward suffers sufficiently suggest supplied sweep theory transform trend trends trio tuning unchanged uniformly usual vendor versus view wavelets white yield yielding yields