Abstract

Three algorithms for solving linearized systems of RF waveform design equations for calculating accelerated spatially-tailored excitations on parallel excitation MRI systems are presented. Their artifact levels, computational speed, and RF peak and root-mean-square (RMS) voltages are analyzed. An SVD-based inversion method is compared with conjugate gradient least squares (CGLS) and least squares QR (LSQR), two iterative algorithms designed to solve large linear systems. The excitation pulses calculated using these methods are used in both Bloch simulations and imaging experiments on an actual eight-channel parallel excitation coil array implemented on a 3T human scanner. Specifically, RF waveforms are designed for accelerated 2D spiral k-space trajectories to produce a variety of 2D target excitations and for a 3D spokes trajectory to produce a uniform thin-slice excitation. Overall, these experiments show that waveforms designed using LSQR and CGLS have significantly lower peak and RMS waveform voltages and produce excitations with fewer artifacts than those generated by the SVD-based method. © 2007 Wiley Periodicals, Inc. Concepts Magn Reson Part B (Magn Reson Engineering) 31B: 176–190, 2007