Advanced software and hardware control methods for improved static and dynamic $B_{0}$ shimming in magnetic resonance imaging

Magnetic resonance imaging (MRI) is a non-invasive tomographic imaging technique and a powerful tool applied in many fields of medicine and research. Methodological developments have vastly broadened the spectrum of possible applications and turned the classical MR scanner into an inherently multi-modal imaging device. Simultaneous hardware advancements have been striving for obtaining better data quality at reduced acquisition times. Here, a central role is taken by the improvement of the MRI magnets, which are required to generate a strong, homogeneous and temporally stable static magnetic field. Any perfectly homogeneous magnetic field, however, is inevitably distorted by the unique magnetic susceptibility distribution of an examination subject. Thus, subject-specific magnetic field shimming technology, which homogenizes the magnetic field, is indispensable and still forms an integral part of current MR research. However, this is non-trivial, because human MR acquisitions are faced with complex distortion field patterns and, thus, many of today’s applications still suffer from strong uncompensated inhomogeneities. Consequently, it was the purpose of this work to identify and overcome existing challenges in subject-specific static magnetic field shimming with a focus on human brain acquisitions. Conventional shim approaches are typically based on 2nd order spherical harmonic shim coils, whose driving currents are adjusted so as to correct for previously measured field inhomogeneities. To gain full control over this process, a comprehensive field mapping and shim current optimization framework was implemented. Accurate shim field characterizations were performed and included into custom-written software. Conducted simulations as well as phantom and in vivo measurements showed, that the shimming framework reaches optimal field homogeneity over user-defined volumes within one minute and without the need for iterations. This way, an average whole-brain field homogeneity of 18.4 ± 2.5 Hz was achieved at a main field strength of 3 T. The shim quality was further improved via the application of a shim coil insert, which can generate very high-order spherical harmonic shim fields. Means of communication with its shim controller were implemented and integrated into the shimming software. Simulations, indicating an improve in whole-brain field homogeneity to 15.1 ± 3.8 Hz, were confirmed in experiments. This residual inhomogeneity was identified as being beyond the correction capabilities even of the high-order shim set and indicates the feasible limit of static spherical harmonic shimming. In contrast, dynamic shim updates during data acquisition can further improve the achievable B0 homogeneity. However, rapid shim changes evoke eddy current-induced distortion fields and, thus, require pre-emphasis corrections. For this, an image-based measurement scheme was developed, which is applicable to capture full 4D eddy current field evolutions. Requiring 10 min of acquisition time, a six-fold acceleration compared to an existing alternative was achieved. Combined with a novel model-based pre-emphasis reconstruction, the shim-induced eddy currents were effectively suppressed and enabled a dynamic operation of the shim hardware. Based on this, a key finding of this work is that the dynamic shim currents and their temporal variation can be strongly constrained while negligibly compromising achievable field homogeneity. Incorporation of these constraints into the shim optimization led to a 23-fold reduction of average maximum and an 18-fold reduction of the average mean inter-slice current changes. Nonetheless, a whole-brain field homogeneity of 8.76 ± 0.32 Hz was achieved, as compared to 8.10 ± 0.31 Hz for the unconstrained case. The associated benefits are manifold, including an intrinsic eddy current reduction, decreased power supply demands and more accurate shim simulations. In conclusion, novel and efficient means to increase the magnetic field homogeneity in MRI measurements were developed within the scope of this work. Especially the methods to measure and reduce shim-induced eddy currents improve dynamic shimming implementations significantly