Sparse magnetic resonance imaging using tagging RF pulses

Abstract

Few fast magnetic resonance (MR) imaging techniques have proposed modifications to the MR signal encoding formulation in order to improve the performance guarantees of image recovery using compressed sensing. A limitation of the previously proposed encoding formulations is their difficult realization on the physical hardware. The deviation of realizable formulation from the theoretical model leads to operating characteristics which are clinically infeasible. In this paper, a novel MR signal encoding formulation using tagging radio-frequency pulses is proposed. The proposed formulation uses tagging pulses to uniquely modulate the longitudinal magnetization in the field-of-view for each MR excitation. The modulation of magnetization leads to mixing of information in the spatial Fourier space which improves the incoherence between the sensing and the sparsifying basis. The physical realization of the proposed formulation is promising due to the use of clinically active RF pulses. The preliminary results for image recovery experiments using the proposed formulation on an in-vivo dataset are comparably close and at times better than the results of the difficult-to-realize state-of-the-art formulation.