Deep learning-based whole-brain B₁ ⁺-mapping at 7T

This study investigates the feasibility of using complex-valued neural networks (NNs) to estimate quantitative transmit magnetic RF field (B₁ ⁺) maps from multi-slice localizer scans with different slice orientations in the human head at 7T, aiming to accelerate subject-specific B₁ ⁺-calibration using parallel transmission (pTx).

Datasets containing channel-wise B₁ ⁺-maps and corresponding multi-slice localizers were acquired in axial, sagittal, and coronal orientation in 15 healthy subjects utilizing an eight-channel pTx transceiver head coil. Training included five-fold cross-validation for four network configurations: $$ used transversal, $$ sagittal, $$ coronal data, and $$ was trained on all slice orientations. The resulting maps were compared to B₁ ⁺-reference scans using different quality metrics. The proposed network was applied in-vivo at 7T in two unseen test subjects using dynamic kt-point pulses.

Predicted B₁ ⁺-maps demonstrated a high similarity with measured B₁ ⁺-maps across multiple orientations. The estimation matched the reference with a mean relative error in the magnitude of (2.70 ± 2.86)% and mean absolute phase difference of (6.70 ± 1.99)° for transversal, (1.82 ± 0.69)% and (4.25 ± 1.62)° for sagittal ( $$ ), as well as (1.33 ± 0.27)% and (2.66 ± 0.60)° for coronal slices ( $$ ) considering brain tissue. $$ trained on all orientations enables a robust prediction of B₁ ⁺-maps across different orientations. Achieving a homogenous excitation over the whole brain for an in-vivo application displayed the approach's feasibility.

This study demonstrates the feasibility of utilizing complex-valued NNs to estimate multi-slice B₁ ⁺-maps in different slice orientations from localizer scans in the human brain at 7T.