The High-Field-Strength Curmudgeon
==================================

* Jeffrey S. Ross

I undertook this assignment, to muse on the state of current 3T clinical imaging, with trepidation because, depending on the phase of the moon, I vacillate between wild unjustified, undocumented optimism of the technology and the opposite extreme of sounding like a fossilized curmudgeon incapable of appreciating technology (or, even worse, of not having the “vision thing”). Our 3T system (*head only*) was installed in June 2001 and runs half-time clinical, half-time research. Because this magnet replaced an aging (yet full-time) clinical system, considerable pressure exists to use it as efficiently as possible for a wide variety of clinical cases. This clinical caseload has not been viewed with excitement by the technologists, who by necessity know the foibles of the individual MR systems more thoroughly than do the neuroradiologists. Their enthusiasm for our 3T system is running slightly short that of spending a night at the Bates Motel.

The possibilities of 3T imaging have captured the imagination of radiologists and the spreadsheets of the manufacturers. The ongoing debate of low versus mid versus high field strength (1.5T) is now elongated to 3T and beyond. What is the reality of current 3T clinical imaging? The operative word is “clinical,” and few sites perform routine protocols, in routine time slots, at this higher field strength for purely clinical reasons. 3T is not presently at the level of mainstream “bread-and-butter” imaging. To undertake a 3T system in a strictly clinical environment begs for, if not trouble, then intense dissatisfaction and a lot of time talking to “Applications.” A clinical system needs to provide excellent image quality quickly and with a minimum of fuss. The concept of good image quality does not refer only to a high-resolution, heavily T2-weighted image but rather the full gamut of T1-weighted spin-echo, balanced, T2-weighted, and more specialized sequences needed in a daily practice. For brain imaging, the ability to produce a high-resolution coronal T2-weighted image of the hippocampus does not validate the complete imaging package at 3 T. The 1.5T platforms have a long history with multiple manufacturers, and the broad expanse of optimized sequences available at 1.5T does not seamlessly transition to the 3T realm.

The hype and promise of 3T relates to the increased signal-to-noise ratio of the higher field strength. Magnetization increases as the square of the field strength, while noise increases linearly giving a potential doubling of signal to noise from 1.5 to 3.0T. This increased signal-to-noise capital can be traded off for higher resolution images at comparable imaging times to 1.5T, better quality perfusion/diffusion studies, or spent for quicker examination times. In reality, this doubling of signal-to-noise ratio has been illusive. Frayne et al did not find the expected doubling of signal-to-noise ratio at 3.0T but an increase on the order of 30–60% (1).

In daily practice, we have not been able to achieve a reduction in overall imaging time because of the constraints of the longer T1 relaxation at the high field and power deposition (2). Producing T2-weighted images is the easy part, and good-quality, high-resolution images are readily obtained at 3.0 T. A reversed T2-weighted image has been encouraged for use on 3T systems as a single encompassing sequence (3). I am loath to give up the multiplicity of sequences that has served us so well, for so long.

T1-weighted images are a much different story. The prolonged T1 at 3T has necessitated the use of either a gradient echo T1-weighted image or an inversion recovery sequence to obtain reasonable gray matter—white matter differentiation. The constraints of power deposition (increasing as the square of the field strength) at the higher field translate into fewer sections with less anatomic coverage. This requires an interleaved sequence, doubling T1-weighted image examination time. T1-weighted image quality is degraded by the increased amount of chemical shift artifact at the higher field. One can obviate this by replacing the conventional T1 with a 3D gradient echo T1-weighted image, but because this is not standard for us at the workhorse field strength of 1.5 T, why should it all of a sudden be acceptable at 3 T?

Limitations also arise in our head-only system because of coil design that distorts the periphery of the images and mandates precise head positioning to achieve optimum signal-to-noise ratio. The images are suboptimal for clinical situations, requiring a high degree of spatial uniformity, such as preoperative stereotactic localization studies, which we still perform on the 1.5T systems. Granted, these problems should be diminished with coil and magnet design improvements and are not such a significant issue on whole-body 3T systems. On the plus side, this inherent problem of the head-only system does relieve one of having to comment on upper cervical degenerative disk disease on the sagittal images, because it is not visible!

Spectroscopy has not yielded a quantum improvement despite the potential of a 100% improvement going from 1.5 to 3.0T, with its improved signal-to-noise ratio and larger chemical shift. Improvements from 20–50% in signal-to-noise ratio can be seen at the higher field strength because of problems related to shortened T2, field inhomogeneities, and increased line width (4, 5).

Not to be a labeled a complete Luddite, immediate improvements can be seen at 3T with nearly any gradient echo technique, such as MR angiography (MRA) and 3D gradient echo sequences. Contrast-enhanced MRA has also been successful demonstrated at 3.0T (6). Susceptibility effects have been surprisingly limited, and diffusion image quality has been very good. Blood oxygenation level–dependent functional imaging studies are excellent, although it is difficult to sell a system to a community hospital on the basis of volume of fMRI studies. In the end, 3T is an evolutionary, not revolutionary, technology. As such, growing pains are to be expected and will be overcome in the future. Five years would be a reasonable time line for full maturation of this technology. In the shorter term, however, I would be cautious about a technology driven by manufacturers and academics when your patient referral base is at stake.

## References

1.  Frayne R, Goodyear BG, Dickhoff P, et al. **Magnetic resonance imaging at 3.0 Tesla: challenges and advantages in clinical neurological imaging.** Invest Radiol 2003;38:385–402
    
    [CrossRef](http://www.ajnr.org/lookup/external-ref?access_num=10.1097/00004424-200307000-00003&link_type=DOI) 
    
    [PubMed](http://www.ajnr.org/lookup/external-ref?access_num=12821852&link_type=MED&atom=%2Fajnr%2F25%2F2%2F168.atom) 

2.  Ruggieri PM, Tkach J, Ross JS, Masaryk TJ. **Routine clinical imaging at 3.0T: an oxymoron?** Radiology 2002;225:430
    
    

3.  Fujii Y, Nakayama N, Nakada T. **High-resolution T2 reversed magnetic resonance imaging on high magnetic field system.** J Neurosurg 1998;89:492–495
    
    [PubMed](http://www.ajnr.org/lookup/external-ref?access_num=9724132&link_type=MED&atom=%2Fajnr%2F25%2F2%2F168.atom) 

4.  Barker PB, Hearshen DO, Boska MD. **Single-voxel proton MRS of the human brain at 1.5T and 3.0T.** Magn Res Med 2001;45:765–769
    
    [CrossRef](http://www.ajnr.org/lookup/external-ref?access_num=10.1002/mrm.1104&link_type=DOI) 
    
    [PubMed](http://www.ajnr.org/lookup/external-ref?access_num=11323802&link_type=MED&atom=%2Fajnr%2F25%2F2%2F168.atom) 
    
    [Web of Science](http://www.ajnr.org/lookup/external-ref?access_num=000168444000006&link_type=ISI) 

5.  Gonen O, Gruber S, Li BS, et al. **Multivoxel 3D proton spectroscopy in the brain at 1.5T versus 3.0T: signal-to-noise ratio and resolution comparison.** AJNR Am J Neuroradiol 2001;22:1727–1731
    
    [Abstract/FREE Full Text](http://www.ajnr.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoiYWpuciI7czo1OiJyZXNpZCI7czo5OiIyMi85LzE3MjciO3M6NDoiYXRvbSI7czoxOToiL2FqbnIvMjUvMi8xNjguYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 

6.  Bernstein MA, Huston J III, Lin C, et al. **High-resolution intracranial and cervical MRA at 3.0T: technical considerations and initial experience.** Magn Reson Med 2001;46:955–962
    
    [CrossRef](http://www.ajnr.org/lookup/external-ref?access_num=10.1002/mrm.1282&link_type=DOI) 
    
    [PubMed](http://www.ajnr.org/lookup/external-ref?access_num=11675648&link_type=MED&atom=%2Fajnr%2F25%2F2%2F168.atom) 
    
    [Web of Science](http://www.ajnr.org/lookup/external-ref?access_num=000171821100016&link_type=ISI) 

*   Copyright © American Society of Neuroradiology