Meeting Banner
Abstract #4201

Limits of Acceleration for Combinations of Compressed Sensing and Parallel Imaging

Ricardo Otazo1, Riccardo Lattanzi1, Daniel K. Sodickson1

1Bernard and Irene Schwartz Center for Biomedical Imaging, NYU School of Medicine, New York, NY, United States

The limits of acceleration for combinations of compressed sensing and parallel imaging remain uncertain. In this work, we investigate the performance of the combined reconstruction with respect to the number of coils for truly-sparse and compressible MR images. A complete basis set of electromagnetic fields is employed as a hypothetical optimal coil array, which achieves the best possible SNR for parallel imaging reconstructions. We demonstrate that the minimum number of required k-space samples is bounded by the number of sparse coefficients, which removes the oversampling factor of 3-5 for compressed sensing alone and approximates the theoretical bounds of L0-norm minimization.

Keywords

acceleration according achieve achieved achieving acquisition acquisitions advantages aliasing allow alone approximates array arrays artifacts assess assuming basis behavior best better biomedical bound bounded brain close coefficients coil coils combination combinations combined complete compressed compute computed conductive considered constraints correlation cylinder cylindrical deciding decreased defined denotes depends distinct elements employed error evaluate even example expansion exploiting fields floor focus free full functions geom gradients grid hods idea ideal incoherence incoherent inefficient infinite intrinsic intro iterative joint know largest like limited limits many maxim maximize meed minimization modeled modes noise nonzero numb optimal orig orthogonal overlapped pace parallel pattern patterns performance planar positioning practical practice process produced pseudo publications pursuit radius random reach reconstruction reconstructions removes require required root sample samples sensing sensitivities sequential share shared shield sign simulated simulations since soft space sparse sparsity square surface surrounded theory though trans transversal trends truly ultimate ultimately umber uncertain useful valuate wave words