Improved Accuracy of Multidimensional Spatially-Selective Saturation RF Pulses for Short-Repetition-Time Steady-State Imaging.

PURPOSE: To introduce and evaluate a novel RF pulse design algorithm for multi-dimensional spatially tailored excitation that is optimized for spoiled steady-state gradient echo MRI sequences. METHODS: Existing multi-dimensional RF pulse design algorithms optimize RF and gradient waveforms using a loss function based on a single RF pulse application. These designs are poorly suited for steady-state MRI sequences where RF pulses are spaced only tens of milliseconds apart, causing potentially large differences between the steady-state magnetization and the target pattern. We propose a multidimensional RF pulse design algorithm with a loss function based on the steady-state magnetization of a spoiled gradient-echo sequence. Our algorithm, steady-state AutoDiffPulses (SS-ADP), extends the previously introduced ADP framework for joint RF and gradient pulse design. We evaluated design accuracy and RF energy in phantom and in vivo brain imaging in inner-volume (IV) and outer-volume (OV) saturation experiments. We further evaluated the potential of SS-ADP to improve reduced FOV (rFOV) imaging. RESULTS: SS-ADP produced significantly more accurate steady-state magnetization patterns compared to a conventional single-excitation design (ADP), with significantly reduced RF energy, for TR ≲ 0.2 T 1 . Furthermore, utilizing SS-ADP for outer-volume suppression (OVS) substantially reduced reconstruction error in accelerated rFOV imaging compared to conventional ADP-OVS. CONCLUSION: A joint RF and gradient pulse design algorithm accounting for steady-state magnetization evolution significantly improves excitation accuracy and reduces SAR in short-TR spoiled imaging with TR ≲ 0.2 T 1 .