Improving GRAPPA using Simultaneous Sparsity

GRAPPA is an effective and commonly-used method for forming reconstructions in accelerated parallel MRI when highly-accurate receive coil field maps (B1-) are not available. This talk explores two approaches to improving GRAPPA image reconstructions at high accelerations, where GRAPPA alone is not adequate. Sparse Reconstruction of Images using the Nullspace Method and GRAPPA (SpRING) is a post-processing method that effectively denoises a GRAPPA reconstruction by jointly optimizing fidelity to GRAPPA and wavelet-domain simultaneous sparsity of the coil images. A second method incorporates simultaneous sparsity regularization into the GRAPPA kernel calibration step, improving reconstruction quality when less auto-calibration data is available. Both methods allow uniform undersampling and thus are a less-radical departure from established image acquisition techniques than some compressed sensing methods.

This talk includes joint work with Vivek Goyal (MIT), Elfar Adalsteinsson (MIT), Leo Grady (Siemens), Jonathan R. Polimeni (Harvard/MGH), and Lawrence L. Wald (Harvard/MGH).

Daniel Weller received his B.S. in Electrical and Computer Engineering with honors from Carnegie Mellon University in 2006, and his S.M. in Electrical Engineering from the Massachusetts Institute of Technology (MIT) in 2008. He is currently affiliated with the Signal Transformation and Information Representation group and is pursuing his Ph.D. in Electrical Engineering at MIT. Daniel is supported by a National Science Foundation Graduate Research Fellowship, and he was the recipient of a National Defense Science and Engineering Graduate (NDSEG) Fellowship. His research interests include signal processing, medical imaging, estimation theory, nonideal sampling and reconstruction, and video and image processing.