MRI techniques such as quantitative imaging and parallel transmit require precise knowledge of the radio-frequency transmit field (B(1) (+)). Three published methods were optimized for robust B(1) (+) mapping at 3T in the human brain: three-dimensional (3D) actual flip angle imaging (AFI), 3D echo-planar imaging (EPI), and two-dimensional (2D) stimulated echo acquisition mode (STEAM). We performed a comprehensive comparison of the methods, focusing on artifacts, reproducibility, and accuracy compared to a reference 2D double angle method. For the 3D AFI method, the addition of flow-compensated gradients for diffusion damping reduced the level of physiological artifacts and improved spoiling of transverse coherences. Correction of susceptibility-induced artifacts alleviated image distortions and improved the accuracy of the 3D EPI imaging method. For the 2D STEAM method, averaging over multiple acquisitions reduced the impact of physiological noise and a new calibration method enhanced the accuracy of the B(1) (+) maps. After optimization, all methods yielded low noise B(1) (+) maps (below 2 percentage units), of the nominal flip angle value (p.u.) with a systematic bias less than 5 p.u. units. Full brain coverage was obtained in less than 5 min. The 3D AFI method required minimal postprocessing and showed little sensitivity to off-resonance and physiological effects. The 3D EPI method showed the highest level of reproducibility. The 2D STEAM method was the most time-efficient technique.
MRI techniques such as quantitative imaging and parallel transmit require precise knowledge of the radio-frequency transmit field (B(1) (+)). Three published methods were optimized for robust B(1) (+) mapping at 3T in the human brain: three-dimensional (3D) actual flip angle imaging (AFI), 3D echo-planar imaging (EPI), and two-dimensional (2D) stimulated echo acquisition mode (STEAM). We performed a comprehensive comparison of the methods, focusing on artifacts, reproducibility, and accuracy compared to a reference 2D double angle method. For the 3D AFI method, the addition of flow-compensated gradients for diffusion damping reduced the level of physiological artifacts and improved spoiling of transverse coherences. Correction of susceptibility-induced artifacts alleviated image distortions and improved the accuracy of the 3D EPI imaging method. For the 2D STEAM method, averaging over multiple acquisitions reduced the impact of physiological noise and a new calibration method enhanced the accuracy of the B(1) (+) maps. After optimization, all methods yielded low noise B(1) (+) maps (below 2 percentage units), of the nominal flip angle value (p.u.) with a systematic bias less than 5 p.u. units. Full brain coverage was obtained in less than 5 min. The 3D AFI method required minimal postprocessing and showed little sensitivity to off-resonance and physiological effects. The 3D EPI method showed the highest level of reproducibility. The 2D STEAM method was the most time-efficient technique.