Accurate MRF-Based 3D Multi-Channel B₁ ⁺ Mapping in the Human Body at 7 T

This work proposes a 3D multi-transmit channel B₁ ⁺ mapping approach based on magnetic resonance fingerprinting (MRF) for the human abdomen at 7 T. A stack-of-stars acquisition is employed to achieve motion-robust 3D encoding, along with a hybrid method where transmit (Tx) channel-wise B₁ ⁺ information is obtained through low flip angle GRE images. B₁ ⁺ mapping at ultra-high field (UHF) in the human abdomen is particularly challenging due to the large dynamic range of B₁ ⁺, the extensive field of view (FOV), and the effects of respiratory motion. Few methods have been proposed to address these challenges, with a significant limitation being the relatively low RF power available at UHF, especially for pTx systems with a 8 × 1 kW power configuration. This limitation makes it difficult to achieve FAs greater than 30° in central body regions, which are required for accurate results with classical methods. In contrast, Tx channel-combined MRF-based B₁ ⁺ mapping has been validated as accurate for FAs greater than 6°, offering improved accuracy at low FAs. Here, two Tx channel-combined MRF-based B₁ ⁺ maps (B1-MRF) are acquired using two tailored complementary phase shims to obtain absolute B₁ ⁺ information across the entire FOV. The 3D hybrid approach was validated against a 2D reference using phantoms and in vivo free-breathing scans in three subjects with varying BMIs, where only one Tx channel was active at a time. The comparison showed strong agreement, with the 3D hybrid acquisition demonstrating improved performance in regions affected by flow, low FAs, or low signal-to-noise ratio compared to the 2D implementation. The higher accuracy and level of detail provided by the proposed method, in contrast to existing methods, are particularly relevant for several applications, including the validation of faster approaches, validation of electromagnetic simulations (which are safety-critical), and the creation of B₁ ⁺ map libraries for applications such as AI-based B₁ ⁺ mapping or universal pulse calculations.