Correcting motion-related B0 Inhomogeneities via Combined Spherical Harmonics and AC/DC Matrix Coils Using AI-based Prediction

Motivation: High-quality MR imaging requires a homogeneous magnetic field (B0) throughout the whole acquisition; however, it can be impaired by motion-related changes. Goal(s): i) Evaluate the effectiveness of static and dynamic shimming in the presence of motion, when integrating spherical harmonics (SPH) with AC/DC matrix coils. ii) Correcting motion-related B0 inhomogeneities using AI-based prediction. Approach: A deep learning network is trained to predict motion-related B0 inhomogeneities and derive correction terms for dynamic shimming. Results: Dynamic shimming was shown crucial to ensure consistent B0 homogeneity when using AC/DC shim coils. Shimming based on AI-predictions effectively mitigated motion-related B0 inhomogeneities. Impact: Dynamic shimming with spherical harmonics and AC/DC matrix coils based on AI-prediction of motion-related B0 inhomogeneities is shown feasible. This offers a promising approach for dynamic shimming that could potentially replace volumetric navigators while maintaining motion-related B0 stability.