B0 shimming, when done well across the entire brain, has the potential to improve many applications including spectroscopy and other quantitative mapping approaches. Dr. Schwerter’s work focuses on hardware driven approaches to improve the B0 shim quality. He has developed several novel hardware control approaches on the basis of very high order spherical harmonic shim coils, which integrate with existing MRI scanners to improve image quality.

What issues in neuroscience or in the clinic makes local B0 shimming a necessity?

I think it makes sense first to define the term “local shimming”, because I think there is a bit of ambiguity here. In the literature, local shimming sometimes refers to homogenizing a static magnetic field over a small region [1], as in single voxel spectroscopy. But local shimming sometimes also refers to additional hardware [2] in the form of small and local non-spherical harmonic shim loops that are placed close to the brain to improve the shim quality for both local small-volume applications and global whole-brain applications.

These advanced shim approaches are most generally a necessity when we have very strong susceptibility gradients like for instance the air tissue interface in the brain. It is very difficult to get an acceptable B0 shim in some of these regions, including the frontal lobe or tissue near the auditory canals. More advanced or local shim approaches are certainly a necessity here. Many frequently used applications would benefit from a more homogeneous field in these regions, from good-old EPI-based methods to novel quantitative mapping approaches. The latter are often particularly prone to errors in the B0, such as recent developments especially in quantitative MRSI, as they all require a more homogeneous magnetic field.

[1] Hetherington HP, Chu W-J, Gonen O, Pan JW. Robust Fully Automated Shimming of the Human Brain for High-Field 1H Spectroscopic Imaging. Magnetic Resonance in Medicine. 2006 Jul;56(1):26–33.

[2] Stockmann JP, Witzel T, Keil B, Polimeni JR, Mareyam A, LaPierre C, et al. A 32-Channel Combined RF and B0 Shim Array for 3T Brain Imaging. Magnetic Resonance in Medicine. 2016;75(1):441–51.

Are there any specific clinical applications that are really being impacted by the lack of good shimming?

I think the vast majority of the of the clinical applications at least at this moment are still qualitative applications that have been used over years or decades even. I think the greatest opportunity is of course in applications that are on the edge of being transferred to the clinic or at least we hope are on the edge of being transferred there. One example here again is spectroscopy, which can provide great additional information for the clinician I believe. But still it is not really used in clinics. Part of the problem is B0 homogeneity. Another example is CEST applications. I find the idea and the physical basis of these concepts very exciting, but the sensitive dependence on many acquisition parameters makes me hesitate as to when these methods will find wide application. And again, the stability and homogeneity of the B0 field plays a role in the robustness of these applications.

How much do shims need to improve compared to the current spherical harmonic model to enable various applications? Is the spherical harmonic model good enough or should we consider alternatives?

I believe that we have seen the potential of the spherical harmonic model, for instance, Hoby Hetherington and his group have shown quite nice results and demonstrated what you can achieve even with the classical spherical harmonic model or with models that go beyond the classic second order [3]. But still we see residuals (unshimmed areas) in brain regions that are beyond the correction capabilities of these systems, and that is where other models may come into play. I think we have seen interesting work from different sites, where people employ local coils to generate correction fields in addition to the 2^nd order scanner shim. I think these methods are currently in a stage in which they show some added value, but still I have not seen a solution capable of homogenizing the field completely and over the whole brain.

[3] Pan JW, Lo K-M, Hetherington HP. Role of Very High Order and Degree B0 Shimming for Spectroscopic Imaging of the Human Brain at 7 Tesla. Magnetic Resonance in Medicine. 2012 Oct;68(4):1007–17.

What techniques do researchers and clinicians use right now to cope with sub optimal shimming?

Often times researchers solve these problems in a certain divide and conquer approach. The large volume gets divided into multiple sub-volumes and then one tries to optimize the field over each sub-volume. Updating the shim during the scan then leads to an overall improvement in shim quality for the entire volume. This can be done over multiple slices in a multi slice experiment or over multiple voxels in multi voxel spectroscopy applications for example.

That is one way to deal with these problems and you can achieve a significant improvement in shim quality. This comes at a certain price, because it takes a bit more effort to calculate appropriate shim currents and to handle eddy currents. But here in Jülich we have shown that this effort can be kept reasonably low with adept methods used for calibration [4] and shim optimization [5].

[4] Schwerter M, Zimmermann M, Felder J, Shah NJ. Efficient eddy current characterization using a 2D image-based sampling scheme and a model-based fitting approach. Magnetic Resonance in Medicine. 2021;85(5):2892–2903.

[5] Schwerter M, Hetherington HP, Moon CH, Pan JW, Felder J, Tellmann L, et al. Interslice current change constrained B0 shim optimization for accurate high-order dynamic shim updating with strongly reduced eddy currents. Magnetic Resonance in Medicine. 2019;82(1):263–75.

What exactly does a local shim coil do to help solve sub-optimal shimming?

The idea is to extend the capabilities of the classical spherical harmonics using hardware that is adapted to human anatomy and the subject you are currently imaging. There is a nice pictorial description in one of the early papers from Christoph Juchem [6]. He had shown that when you place a local shim coil close to the face, you can generate a local field that would ideally cancel out the inhomogeneity in the frontal lobe caused by the nasal cavity. And then you place a few additional correction loops to fully shim all of the difficult shim regions. That is the basic idea.

This concept has been extended over the last couple of years where researchers have used more complicated local shim coil setups with multiple loops arranged also sometimes in different shapes over a cylindrical surface and thereby trying to generate a correction field that is ultimately adjusted to compensate for all local inhomogeneities.

[6] Juchem C, Nixon TW, McIntyre S, Rothman DL, de Graaf RA. Magnetic Field Homogenization of the Human Prefrontal Cortex With a Set of Localized Electrical Coils. Magnetic Resonance in Medicine. 2009;NA-NA.

How does this additional hardware get integrated into the scanner?

Fortunately, the shimming process can be seen mostly as a standalone process because one can assume to a reasonable degree that the fields that you generate in the shim hardware is not interfering with the hardware or the imaging process of the scanner itself.

You can acquire a regular field map, decompose it into the components that you can correct with your shim system and then update the currents in the shim coils. Hardware integration, power supply communication and transfer of data for shim processing are therefore relatively straightforward.

What effect does all this extra hardware have on subject handling? Overall, how do you see this affecting the overall subject experience?

The value that the subject receives potentially from an improved shim certainly is that the clinician can get access to different information from more advanced techniques which give more specific information on the patient condition, particularly for the quantitative imaging parts. If these improved shim techniques enable new sequences to be played out in clinics on a broader basis, that certainly is a big benefit for all patients.

In general, the patient experience, if we are extrapolating from what we have right now is probably only a very slightly prolonged scan preparation time because we have to set up the additional shims or configure new shim methods. However, this could allow us to use applications that cannot otherwise be used due to poor and homogeneity. It’s really worth the effort. And there has been a nice comment on that recently published by Chris Wiggins [7]. He comments on the fact that there have been little advances in shimming over the past decades and that we should still invest more time or resources to come up with ideas and solutions for the shimming that is acceptable in the clinics in terms of patient handling and time in the scanner. I do really share that opinion.

[7] Wiggins CJ, Choi C, Li Y, Lin AP, Thakur SB, Ratai EM. Shimming—the forgotten child of in-vivo MR? Magn Reson Mater Phy. 2021 Apr;34(2):179–181.

What have been the greatest challenges in terms of getting these shim systems implemented in research or clinical environments? Has there been any pushback from researchers in terms of the new workflow, or have challenges largely been technical?

I think if it were just a technical problem, we would have seen clinics or research sites that have adopted these techniques on a broad basis. I think the condition that we today still haven’t seen or still do not see these sites is due to the fact that we still do not have the game changing shimming technology. For example, the dynamic shimming methodology can solve the shim problem or at least improve the quality to some degree in slice-by-slice acquisitions, but there are still true 3D applications or other volumetric techniques for which the shim still is insufficient, and which cannot benefit from dynamic shim updating. If we are able to find the solution to this shimming issue, one that provides sufficient homogeneity and is robust and easy to use, then I think everybody would certainly immediately adopt it. For brain imaging, I believe we will need hardware that can more specifically generate that characteristic correction field that is required to simultaneously target all brain regions. Given the fact that collaborative research and open access developments are being pushed over the past few years, I’m optimistic that we will have better shim solutions available for many research sites in the future.