The field-strength war
ike almost everything in this world, MR machines come in different sizes: extra-small, small, medium, large, and extra-large. The technical terms in MR lingo for these sizes are ultralow, low, medium, high, and ultrahigh field machines. These terms refer to the magnetic field strength of the respective machine. The field strength is measured in Tesla (T), a unit that replaced the former unit of Gauss (G) some years ago, although Gauss is still used sometimes (10,000 G = 1 T).
Ultralow-field machines operate at a field strength below 0.1 T, low field between 0.1 and 0.5 T, medium field between 0.5 and 1 T, high field between 1 and 2 T, and ultrahigh field machines above 2 T.
In clinical surroundings, the national radiological protection boards used to allow machines as high as 2.0-2.5 T. Everything above this limit was considered potentially hazardous and thus should only be admitted to research facilities – particularly if fast gradient-switching was used. Today, ultrahigh fields are considered safe for research and, partly, for clinical routine – at least in some countries.
In describing MR machines, natural scientists prefer to talk about frequencies instead of field strengths. This is because different nuclei in the periodic system possess different resonance frequencies. At 1 T, for instance, protons resonate at 42.58 MHz, whereas at the same field strength, phosphorus nuclei resonate at 17.23 MHz. For clinical imaging purposes in medicine, this is of no importance because only proton MR imaging is used.
Strolling down the aisles of the world’s biggest commercial exhibition of medical imaging equipment at the annual meeting of the Radiological Society of North America, one could find small machines operating at 0.06 T and huge machines operating at 4.0 T or even higher fields. Their magnets are different: below approximately 0.3 T, the magnets are permanent and resistive or electromagnetic, but above this field the magnets are superconductive. All these magnet types have their pros and cons.
Why does one find small ultralow field MR imagers and high field machines operating at fields 100 times stronger? Why are there not only low or high field machines?
The field-strength question has divided the MR community since the early 1980s. At that time, all MR machines operated at low fields, and many of the prototypes had strengths of approximately 0.15 T. Researchers did not believe that imaging at higher field would be possible because higher radio frequencies would not be able to penetrate the human body. Like many other predictions in MR imaging, this prediction was wrong.
MR images at that time were crude, blurry, and generally worse than CT images. Scientists working for the R&D divisions of companies producing MR equipment were asked:
“How do you get better image quality?” They had a simple answer: “Increase field strength.”
From analytical applications of chemical MR spectroscopy it was known that the signal-to-noise ratio increases when field strength is increased. The better your signal-to-noise, the better your image will be. Higher fields also require higher gradient strength to reduce the chemical-shift artifacts created by these fields. In turn, this led to better spatial resolution. So some manufacturers, driven by their research and marketing people, moved to high-field superconductive magnet systems. These systems were (and in some instances still are) huge, dinosaurlike machines. They were expensive, difficult to produce, and costly to maintain, but image quality suddenly became better.
Another argument supported the development of high field machines; only these machines are able to produce in vivo MR spectra for phosphorus or proton spectroscopy. At this time, one of the aims in the development of MR in medicine was to combine imaging and spectroscopy to acquire morphological and metabolic information about the human body. The higher the field, the more detailed spectroscopic information will be.
However, in vivo spectroscopy did not take off, whereas the popularity of MR imaging exploded. Dedicated imaging machines became the rule, combined imaging and spectroscopy the exception.
Even for imaging, it became an ideology to plead for high fields. There is no rational scientific reason for this development; image quality and spatial resolution of low and medium field machines became as good as, and in some instances even better than, that of high or ultrahigh field equipment. Additional research revealed that the most important factor in medical imaging, tissue contrast, at least for certain diagnostic questions in the central nervous system, seems to be best at medium fields and, in some instances, even decreases with higher fields [2,3].
There was still no rational approach to the problem. At a 1983 magnetic resonance conference in San Francisco, a debate on field strength that had started on the platform was continued in the corridor of the conference center. The discussion nearly ended in a fist fight between the proponent of the high field ideology, whose company had put all its efforts into 1.5 T machines, and the proponent of low fields, whose company advocated MRI systems at 0.35 T.
The front lines in this war were mighty and the trenches deep. You were either part of one camp or the other. All large companies jumped on the high field side and promoted high fields with all the ammunition their marketing departments could provide. In some countries, millions of dollars of taxpayers’ money were channelled into subsidies for the development of high field systems.
However, one morning in the early 1990s MR customers woke up and found that a gap was emerging. One company had decided to enter the mid-field market, another followed suit, and a third decided to compromise by offering an MR machine operating at a field strength in between the others.
The reasons for these steps were never publicly discussed, but people had realized that the signal-to-noise increase expected from the results in analytical NMR did not occur in the same way in whole-body MR imaging.
In whole-body MR imaging, signal-to-noise increased to a certain extent, and then the human body created additional noise that led to a flattening of the signal-to-noise curve at high fields. In addition, nobody had foreseen the new problems faced by users at higher fields, among them being the worsening of motion and susceptibility artifacts. Cost and hazards also increased with higher fields. At the same time, low and medium field machines became smaller, the quality of their diagnostic output better, and interventional MR became feasible.
A new generation of buyers, the smaller hospitals and private practices, preferred cost-efficient MR systems that they could use for most of the daily routine examinations. Bigger hospitals, and in particular those interested in spectroscopy and research in functional imaging, went for high field systems, but for them also the second and third system usually was medium or low field. Today, the market for 1.5 T high field equipment is nearly unbroken because they are good diagnostic machines.
"Definitions always seem to be in the eye of the beholder."
The marketing department of the biggest US-American manufacturer pushed for high field (1.5 Tesla) in the 1980s. Fifteen years later, they postulated that their new mid-field equipment (0.7 Tesla) was also high field. Another 15 years later, 3 Tesla was the non-plus-ultra and "high field". Definitions always seem to be in the eye of the beholder. If all of this had been known or taken into consideration 15-20 years ago, more patients would have had access to MR imaging, and medical MR equipment might have been less expensive than it is today.
Derek Shaw worked for Varian, later for Oxford Instruments, and since 1983 until his retirement for General Electric Medical Systems. He is one of the leading MR scientists in Europe. In 1996, he wrote the following statement in a book chapter:
“The early period of MRI ... was dominated by the 'field-strength war'. What was the best field strength for MRI? These battles were essentially commercial, science being used to justify the company’s competitive position ...
“Our pawn in the field strength battle was in vivo spectroscopy... As it became apparent that there was not going to be sufficient specificity available via T1 and T2 determinations, MRS ... was seen as a potential alternative ... MRS needed the highest field possible ...
“This need, along with the higher signal-to-noise ratios achievable at higher field strengths ... led, despite their extra costs, to the use of 1.5 T magnets ... Without this push to high field, MRI systems might be quite different today, probably lower down on the cost/performance scale.”
Thus, the trend went towards high field. High field made higher profit, which is a recurrent theme not only in medical technology. This is reflected in equipment sales worldwide and in the sales revenue of MR equipment according to field strength.
Recently the field strength quarrel has flared up again. This time it is 1.5 Tesla versus 3.0 Tesla. However, this time it is not MR spectroscopy, but functional MR imaging pushing up field strength. The results are the same. In research – and in some places even in routine imaging – 3-Tesla system have become the sine qua non. It has become fashionable to buy them.
They also have some advantages over low field equipment; for instance, ultrafast imaging, where scan time is reduced at the expense of signal-to-noise ratio, is generally more effective at higher fields. This facilitates another ‘sexy’ research area: functional imaging of the brain.
Once again, people claim that the signal-to-noise ratio in MR imaging increases linearly with field strength. Some researchers state that signal-to-noise between 1.5 T and 3.0 T increases by 200%, even 300%. There are papers indicating that this might be correct for functional MR imaging using the BOLD technique. However, comparing the BOLD technique and MR imaging is like comparing apples and oranges. On the other hand, for MR spectroscopy a 20% increase in sensitivity, but the same signal-to-noise ratio has been shown in comparative studies between 1.5 Tesla and 3.0 Tesla .
For MR imaging itself, reliable comparative studies do not exist – to make a valuable comparison, the total amount of signal acquired during the same time should be taken into account and this is not in favor of ultahigh fields since T1 increases.
Without any doubt, signal-to-noise and spatial resolution can be better at 3 Tesla, coronary arteries are better seen, small brain structures better delineated. However, the cost/benefit ratio remains unknown ... and adverse effects might threaten both patients and staff.
Next stop: 7 Tesla, perhaps 9.
1. Barker PB, Hearsham DO, Boska MD. Single-voxel proton MRS of the human brain at 1.5 T and 3.0T. Magn Reson Med 2001; 45: 765-769.
2. Hoult DI, Chen C-N, Sank VJ. The field dependence of MRI II. arguments concerning optimal field strength. Magn Reson Med 1986; 3: 730-746.
3. Rinck PA, Fischer HW, Vander Elst L, Van Haverbeke Y, Muller RN. Field-cycling relaxometry: medical applications. Radiology 1988; 168: 843-849.
4. Shaw D. From 5-mm tubes to man. The objects studied by NMR continue to grow. In: Grant DM, Harris RK. Encyclopedia of nuclear magnetic resonance. Volume 1, Historical perspective. Chichester: John Wiley and Sons. 1996, 623-624.