A fieldmap-driven few-channel shim coil design for MRI of the human brain

Abstract

We exploit the inter-subject similarity of inhomogeneous static magnetic field patterns arising in the human brain under MRI examination to design a small set of shim coils providing performance equivalent to numerous coils based on high-order Spherical Harmonics corrections. A hundred brain B 0-maps were first collected at 3 T. Ideal subject-specific electric current density stream functions are then computed with low power constraints, on a cylindrical surface. This step is repeated over tens of brain maps so that a Principal Component Analysis can be applied to the stream functions; the main components result in the small set of coils. Both 50-subject hold-out and 10-fold cross-validation are employed to evaluate consistency of the proposed system performance over a posteriori subjects. Simulations show that only 3 cylindrical coils manage to capture the principal magnetic field profiles in the human brain, thus providing a better static field inhomogeneity mitigation than that obtained from 16 unlimited-power high-order Spherical Harmonics coils, with inhomogeneity greatly reduced in the pre-frontal cortex compared to 2 nd-order shimmed baseline field acquisitions. The approach provides a very reduced channel count system for mitigating complex B 0-inhomogeneity patterns. Thus, a compact, cost-effective system could be conceived and driven by relatively low-budget electronics. The method should therefore have a strong impact in both Ultra-High and portable low-field MRI/MRS. Moreover, this technique can be applied to the design of shim coils addressing anatomies other than the brain.

Physical limits to human brain B0 shimming with spherical harmonics, engineering implications thereof

“…The database was built from fieldmaps acquired on a MAGNETOM Prisma 3 T imager (Siemens Healthcare GmbH, Erlangen, Germany) with 1.7 mm isotropic resolution, as already reported in [13]. Assuming no shimming of the brain would affect the quality of some δB 0 maps with regards to distortion and signal loss, the fieldmaps were acquired subsequent to 2nd-order shimming by the Prisma scanner.…”

“…In order to explore practical aspects of coil design, and how realistic cylindrical systems could perform relatively to the best achievable inhomogeneity, the Dipole Boundary Method (DBM) [13] is applied to compute subject-optimal stream-functions (SO-SFs) for each subject in the database under global and slab-specific shimming techniques.…”

“…As an example, in non-accelerated Echo-Planar Imaging (EPI) single-shot acquisitions, under 0.5 ms inter-echo spacing and 200 mm Field-of-View (FOV) in the phase encoding direction, a 100 Hz excursion in the magnetic field leads to 10 mm geometric distortion in the reconstructed image [3][4][5]. It is therefore not surprising that appreciable effort has been directed to the design of shimming systems for the human brain [8][9][10][11][12][13][14][15], but as we will see, they are still far from achieving the minimal inhomogeneity.…”

“…Moreover, non-SH-based Multi-Coil Array (MCA) systems [8,9,25,26] have gained traction in the last years. These have been shown to provide adequate homogeneity for numerous applications at Ultra High Field (UHF), particularly in dynamic shimming mode [8,10], but strong field excursions persist around the ear canals and in the pre-frontal cortex despite shimming, even when employing brain-optimized MCAs [11][12][13][14][15][27][28][29].…”

Iterative static field map estimation for off‐resonance correction in non‐Cartesian susceptibility weighted imaging

“…17 Although this approach provides a significant improvement on image quality, a complementary technique is required in the most demanding setups such as in EPI 18,19 or in non-Cartesian imaging, that makes use of long readouts. 12,13,[20][21][22] Recent works have been proposed to improve this method using more efficient coil designs 23 that accurately fit the expected inhomogeneity profiles 24 using complementary data. Those experimental results permit to achieve improved correction performances, but remain limited by the number of shimming coils to be deployed to reach the theoretical limits.…”