Physics of high field MRI

In this Educational Lecture, I will discuss the physics and clinical applications of high field MRI. Several advantages can be identified when moving to a higher magnetic field strength:

1. Increased SNR Can be used to acquire higher spatial resolution images within reasonable scan times or to scan faster with a ‘normal’ spatial resolution, and increases sensitivity to x-nuclei like ²³Na and ³¹P^1-4.
2. Increased CNR Caused by changes in T₁ and T₂ relaxation times and increased susceptibility effects); leads to higher image contrast between tissues irrespective of spatial resolution⁵, can be used for better background suppression in TOF MRA^6,7, and enables clearer visualization of tissues and materials with a high magnetic susceptibility^8-10.
3. Larger chemical shift Individual metabolites and macromolecules can be more readily differentiated when using metabolic MRI techniques like CEST and MR spectroscopy¹¹.

However, nothing comes without a cost:

1. Magnetic fields (B₀ and B₁) are inhomogeneous Can cause difficulty imaging near air-containing structures, and/or a spatially varying image contrast within the same slice; (partial) solution = shimming & dielectric pads^12,13.
2. RF energy becomes less homogeneously distributed Causes SAR hotspots and thereby restricts sequence development.
3. Custom-made coils necessary Commercially available coils are limited to a head coil and (recently) a knee coil; for other body parts, coils need to be custom-made^14,15.
4. More contraindications Fear of heating of metallic implants has long restricted patient scanning; new insights and experience have shown that these risks might not be much different from lower field strengths¹⁶.

Clinical applications

Considering the clinical applications of high field MRI, a distinction can be made between developments that have broadened our knowledge of diseases – e.g. their development, associated factors and their effects on other organ systems – versus clear-cut applications that are directly applicable in clinical practice – e.g. by improving diagnosis, assessing treatment response and predicting prognosis. In my lecture, I will focus on the latter.
Several neurological diseases benefit from the use of high field MRI. In approximately 30-40% of MRI-negative epilepsy patients, an epileptogenic lesion can be found with high field MRI, and in patients with already known lesions it can better delineate the lesion, both aiding surgical treatment of these patients^17,18. In a similar fashion, pituitary microadenomas can be detected using high field MRI in patients without a lesion on lower field strengths, and visual pathway damage due to larger lesions can be assessed in higher detail, improving treatment of these patients^19,20. High field MRI has also been successful in difficult clinical cases like diagnosing Parkinson’s disease, differentiating between Parkinsonisms, and it can aid in positioning of deep brain stimulation electrodes, although whether the latter really leads to improved treatment is not known yet^21-24. In MS, initial advantages like visualization of the small central vein that can differentiate from vascular white matter lesions are now also possible using 3T; nowadays, high field MRI can identify patients prone to a more severe disease course through detection of iron rims and cortical lesions which are difficult to see at lower field strengths^25-27. Finally, in brain tumors high field MRI can be used to non-invasively differentiate between IDH-mutated and IDH-wildtype gliomas, detect treatment response and visualize tumor metabolism with MR spectroscopy and glutamate-CEST; however, these applications are still in their early clinical phase^28-30.
Clinical studies outside the brain have been more sparse, and I will highlight some of the more promising developments for clinical practice. Ultrashort echo time imaging, T₂ mapping and sodium MRI have been used successfully for detecting early cartilage lesions in e.g. the knee, and visualizing aging and inflammation^31-33. In the breast, sequences like APT-CEST and ³¹P MR spectroscopy are able to visualize a breast tumor without using a contrast agent, and can assess early treatment response to chemotherapy^34,35. In cardiac imaging, high field MRI can be used to improve the temporal resolution – i.e., faster scanning – while ³¹P MRS has been tentatively used to visualize hypertrophic cardiomyopathy^36,37. Finally, abdominal imaging has been performed at high field to for instance detect prostate cancer, very small pelvic lymph nodes, and subtle parametrial invasion in cervical cancer, using either a body coil or, more frequently, a smaller local coil like the endorectal coil^38,39. However, these developments, like those in the heart, are still in a very preliminary phase.
Several open questions remain, including (1) Is high field MRI really better than lower field strengths, i.e. does it really have an added value for patient care? Current opinions are scattered, and for almost all clinical applications including MS and epilepsy^40-42. (2) Is high field MRI necessary for everyone, or just for a selected few? (3) When have we reached the limit in field strength? Most clinical applications are performed at 7T; what would even higher fields bring us?^43,44 These are some of the questions that we will have to focus on in the near future.

Acknowledgements

A big thank you to the 7 tesla MRI group at the UMC Utrecht for their contributions.

References

¹Kraff et al. (2015) JMRI 41:13-33; ²Nagel et al. (2014) Radiology 271:585-595; ³Biller et al. (2016) Sci Rep 6:31269; ⁴Umathum et al. (2013) Radiology 269:569-576; ⁵Zwanenburg et al. (2010) Eur Radiol 20:915-922; ⁶Harteveld et al. (2015) PLoS ONE 10:e0121051; ⁷Zhang et al. (2015) MRI 33:1043-1050; ⁸Ropele et al. (2011) JMRI 34:13-215; ⁹Theysohn et al. (2011) JMRI 33:782-791; ¹⁰Ladd et al. (2018) Prog Nucl Magn Reson Spectrosc 109:1-50; ¹¹Godlewska et al. (2017) Front Psychiatry 8:123; ¹²Teeuwisse et al. (2012) MRM 67:1285-1293; ¹³Stockmann et al. (2018) NeuroImage 168:71-87; ¹⁴Santini et al. (2018) MRI 45:1-6; ¹⁵Platt et al. (2018) MRM 80:1005-1019; ¹⁶Schick et al. (2005) Eur Radiol 15:946-959; ¹⁷Wang et al. (2020) Epilepsia 61:2509-2520; ¹⁸Feldman et al. (2019) PLoS ONE 14:e0213642; ¹⁹Patel et al. (2020) Neuroradiology 62:1051-1054; ²⁰Rutland et al. (2019) J Neurosurg 132:333-342; ²¹Maruyama et al. (2019) Sci Rep 9:18357; ²²Bot et al. (2019) Stereotact Funct Neurosurg 97:153-159; ²³Xiao et al. (2016) Front Neurosci 10:264; ²⁴Frosini et al. (2016) J Neural Transm 23:509-513; ²⁵Dal-Bianco et al. (2021) Brain 144:833-847; ²⁶Tallantyre et al. (2011) Neurology 76:534; ²⁷Treaba et al. (2019) Radiology 291(3):740-749; ²⁸Xu et al. (2015) Tomography 1:105-114; ²⁹An et al. (2018) MRM 79:1851-1861; ³⁰Grabner et al. (2017) Eur Radiol 27:1556-1567; ³¹Zbyn et al (2020) Invest Radiol 55:430-437; ³²Menon et al (2021) Magn Reson Imaging Clin N Am;29:117-127; ³³Lazik-Palm et al (2020) Eur Radiol Exp 4:10; ³⁴Loi et al (2020) Eur J Radiol 129:109068; ³⁵Krikken et al (2019) NMR Biomed e4086; ³⁶Ibrahim et al. (2020) World J Radiol 12:231-246; ³⁷Valkovic et al. (2019) J Cardiovasc Magn Reson 21:19; ³⁸Hoogendam et al. (2017) Eur Radiol 27:938-945; ³⁹Fortuin et al. (2020) PLoS ONE 15:e0236884; ⁴⁰Wang et al. (2021) Front Neurol 12:591586; ⁴¹Gonzalez-Escamilla et al. (2021) Mult Scler 27(3):362-363; ⁴²Louapre et al. (2021) Mult Scler 27(3):360-362; ⁴³Geldschlager et al. (2021) Magn Reson Med 85:1013-1027; ⁴⁴Ke et al. (2020) Hum Brain Mapp 41:4529-4548