GRAPPA operator for wider radial bands (GROWL) with optimally regularized self‐calibration

Abstract

A self-calibrated parallel imaging reconstruction method is proposed for azimuthally undersampled radial dataset. A generalized auto-calibrating partially parallel acquisition (GRAPPA) operator is used to widen each radial view into a band consisting of several parallel lines, followed by a standard regridding procedure. Self-calibration is achieved by regridding the central k-space region, where Nyquist criterion is satisfied, to a rotated Cartesian grid. During the calibration process, an optimal Tikhonov regularization factor is introduced to reduce the error caused by the small k-space area of the self-calibration region. The method was applied to phantom and in vivo datasets acquired with an eight-element coil array, using 32-64 radial views with 256 readout samples. When compared with previous radial parallel imaging techniques, GRAPPA operator for wider radial bands (GROWL) provides a significant speed advantage since calibration is carried out using the fully sampled k-space center. A further advantage of GROWL is its applicability to arbitrary-view angle ordering schemes.