Some license can be allowed to the old, permitting them a measure of reminiscence, particularly when memories of important times and events are beginning to fade. I remember standing with Paul Lauterbur in the conference hall during the Sydney ISMRM meeting in 1998, when he said, “You know, many of the people here weren't born when I published my paper in Nature in 1973.” If that statement were true in 1998, it is even more true today; and every day, as more and more people start work on MRI, the truth of Paul's remark is further reinforced. It is permissible, then, for one who was not there at the very start (but arrived very shortly after) to attempt to convey a little of the excitement and uncertainty of the early years. I would also like to attempt to set the importance of the work that led to the award of the Nobel Prize to Paul Lauterbur and Peter Mansfield this year into the framework established by the pressures and ambitions of the time it was done and, in passing, to support the correctness of the Nobel Committee's assessment of what was important, and what was largely irrelevant, to MRI as we know it. NMR itself had a somewhat tortuous genesis. Bloch and Purcell, who received the Nobel Prize for its discovery, were not the first to attempt to find the resonance that had been predicted over a decade before their demonstration of it. Gorter, a very distinguished Dutch physicist, published his first attempt in 1936. Unfortunately for him, the materials he had used as samples were completely inappropriate. Biologic investigations began immediately after the first demonstration of the resonance, when Bloch inserted his finger into his magnet and saw a large resonance signal. After the pioneering studies that led to the understanding of NMR relaxation mechanisms, the first group to study the relaxation behavior of a wide variety of human tissues (albeit in vitro) was led by Odeblad (who began his studies with Purcell's group in Harvard, but later continued them in Stockholm). Curiously, while many tissues were studied by this group, they did not look at cancerous tissue. This was done some 10 years later by Raymond Damadian, who subsequently claimed that the increased values for T1 and T2, which he had observed in vitro in a mouse tumor model, would be diagnostic. As is now well known, a huge variety of pathologic processes result in increases in the relaxation time constants, while some classes of tumor have shorter time constants than the normal tissues from which they have developed. Sadly, the many attempts that were made to correlate pathology and relaxation behavior have yielded none of the precise numerical relationships that were hoped for in the early days of MRI, so that this line of investigation, which was studied extensively in the early days of imaging, has now been almost completely abandoned. Raymond Damadian also produced a sketch of a possible NMR imaging system in a patent filed in 1972. The method he first proposed involved quite impracticable concepts and could never have been implemented. When he did produce an imaging system in 1977 (the so-called “indomitable” machine), it was quite different and used a more primitive version of the “single-point” acquisition technique that Hinshaw and his colleagues at Nottingham had already described. Damadian used this machine to generate an image of the human thorax in 1977. However, the approach, dependent as it was on the excitation of only one voxel at a time, and its subsequent detection, was extremely slow. The method was, unfortunately for Damadian, one of those classic blind alleys that lead precisely nowhere. Donald Hollis, in his book Abusing Cancer Science: The Truth About NMR and Cancer, published in 1987, has a great deal more to say, little of it complimentary to Damadian, about the various claims he has made about both cancer diagnosis and imaging. The key discovery needed to turn NMR imaging into the clinical reality that it has become was that described in Paul Lauterbur's Nature paper in 1973. Subsequently, Peter Mansfield led those who began the process of converting Paul's work into a medical tool. Interestingly, MRI (as it became later, after concerns about a group of protesters in Cleveland, Ohio, who became rather excited about the word “nuclear” in the original description) was only one of a clutch of imaging and diagnostic methods that were first published between about 1972 and 1976—possibly as a result of the excitement engendered by the introduction of Hounsfield's CT scanner in 1971. For the first time, it seemed to scientists, engineers, and doctors that computers might have a genuine role in medicine, permitting the introduction of procedures that simply were not possible without them. PET and MRS, as well as MRI, optical imaging (including near infra-red studies), microwave imaging, and magnetoencephalography all had their beginnings in these few years. Every part of the electromagnetic spectrum was considered as a potential target for imaging investigations, with particular attention being paid to parts of it in which the body was more or less transparent to the radiation. Radiofrequencies are one such window. Tomography in nuclear medicine was roughly coincidental with that in x-rays, though much less often recognized, while, at the time, ultrasound was an essentially analogy procedure, with Doppler ultrasound only appearing later in the decade. Of the new introductions, though PET devotees may disagree, only MRI has emerged as a major modality. And this is wholly and entirely due to the use of gradient fields. And it was for the development of the use of these that this year's Nobel Prize for Medicine was given. By modern standards, particularly those of the computer industry, it took a long time for MRI to arrive. In 1974, Aberdeen University unintentionally demonstrated the method's ability to show pathology when they had to kill a mouse they were imaging in order to have it still enough for the approach they were using at the time to produce an acceptable image; they then saw the signal changes due to the edema that formed at the point at which its neck had been broken. Various objects and parts of the body were imaged by a variety of techniques during the years up to 1977 and the first image of the human thorax. The first image of the head was not acquired until 1978 and was taken, interestingly only in the historic context, by a technique exactly analogous to Hounsfield's original “step and rotate” CT method. There was quite intense rivalry amongst the various groups to achieve notable “firsts,” and it fell to those at Nottingham to propose how to acquire images quickly (one of the achievements for which Peter Mansfield was cited) and demonstrate how sagittal and coronal images could be acquired directly, while the Aberdeen group developed Kumar, Welti, and Ernst's proposed, though barely practical, method for obtaining two-dimensional images into the spin warp method, which is the basis of the vast majority of all the images acquired today. Meanwhile, Paul Lauterbur had laid the foundations for the design and use of modern MRI contrast agents, and the Hammersmith group showed how to control image contrast. In the early 1980s, the group at Diasonics/UCSF completed the basic repertoire of imaging capabilities with the introduction of the multi-echo, multi-slice approach that made MRI financially viable in clinical terms, following through from the first clinical series that had been performed in Aberdeen in 1980. The first genuine production machines did not reach the market until 1982, and, even then, these were supplied by smaller companies usually with little or no CT market share to lose. The major diagnostic equipment companies did not begin to deliver machines until 1985/6, when they began to ship the 1.5-T systems that have turned MRI into the examination of choice across so much of diagnostic medicine. The gestation period for MRI can genuinely be said to have been 12 or so years, comparable with that of a drug. Interestingly, again, it was largely a quirk of history that led to MRI systems taking the form they have today. The Nottingham and Aberdeen groups both saw MRI as a cheaper alternative to CT, which, though developed in Britain, was deemed too expensive for the British National Health Service, after the initial small batch of machines had been bought for Regional Neurological Centers. Waldo Hinshaw and Bill Moore (after whom the Physics Young Investigators Award prize is named) concentrated on SSFP, and their design—based on a resistive magnet, as was everyone else's then—was genuinely cheap. It suffered from the one disadvantage then—which nowadays would be a problem that would soon be resolved—that, with its signal dependency on T1/T2, it had very poor soft tissue contrast. The Aberdeen group, in turn, developed spin warp because it was much more tolerant of bad B0 fields than most approaches, thus, again, leading to potentially cheap systems. The Hammersmith group, working with EMI, had originally planned to use DEFT, but this proved impossibly demanding with the equipment available, and it switched to pulse and collect in a variety of forms, which did allow for true contrast manipulation. Under pressure from the Department of Health, who were partially funding the work and who insisted on the incorporation of Peter Mansfield's ideas on fast imaging, the group went to a cryogenic magnet designed to work at 0.3 T to improve the system's signal-to-noise ratio. The system that was developed was subsequently run at 0.15 T for reasons beyond the scope of this editorial, though they were not of a technical nature. At the time, the group were regarded as traitors to the cause of MRI, who had prostituted the purity of the ideals governing its development for unworthy commercial ends. As can be seen from the foregoing, there are direct analogies between the initial discovery of NMR and that of MRI. In both cases, there were pioneers who pointed out that there might be a way forward; in both cases, these failed to successfully implement what they had conceived. Subsequently, in both instances, the Nobel Prize was awarded to those who had discovered what ultimately turned into a highly valuable technique and provided the framework that could be exploited by others. In the case of MRI, the award was given, explicitly, for the development of MRI using spatial location by gradients. No one can rationally argue that this has not proved to be the only useful approach and can, therefore, challenge the legitimacy of the 2003 award. One thing that has not been commented on elsewhere, as far as I know, is how important it has been that MRI was discovered when it was. There is a very good case for arguing that had it been proposed either 20 years before it was (which could have been possible) or 20 years later, it would never have achieved other than a curiosity value. The Fourier transform technique was first proposed in the early-mid 1950s for use in infrared spectroscopy (though only published by its discoverer in his Cambridge PhD thesis, by a curious historic oversight). Before the introduction of the minicomputer, it was too cumbersome and awkward, so that Ernst and Anderson in 1966 could usefully apply it to NMR, though doing so much earlier would not have been productive. Gabillard's studies using gradients in 1951–1952 were, in effect, one-dimensional imaging ones, though they led to nothing remotely resembling modern imaging. On the other hand, it must be questioned as to whether, if Lauterbur's paper had been published in 1993, all other technologies having evolved as they have, MRI would have been anything more than a scientific curiosity of interest to a group of physicists for a period, before being consigned to history as largely irrelevant. Four factors would have militated against its success. First, the research funding agencies in the English-speaking world, in particular, are obsessed by genetic research. Unfortunately, funding of biophysics and bioengineering (both of which have, I believe, demonstrated much better cost effectiveness than the huge sums poured into genomics in its broadest sense) is very limited, and the sort of sums needed to create a useful research system are very hard to come by these days. Second, after a further 20 years of CT development, the equipment industry would find it even harder than it did in the 1970s to accept a new unproved technology. EMI took the view when it began work on NMR imaging (as it then was) that it would take 9–10 years to bring a suitable product to market. It was still prepared to invest, but it is unlikely that many industrial companies would do so now when faced with this kind of timescale. Interestingly, at the time when the major instrument companies proceeded very cautiously, the industrial lead (apart from EMI and Damadian with his FONAR company) was largely taken by what were primarily drug companies. Third, the radiologic profession, again with 20 more years of experience with CT and using much more effective modern machines, would regard the additional expense and training needed for a new modality with much more skepticism than it did in the early 1980s at a time when the euphoria surrounding computerized imaging was still at its height. It would now require much more convincing before adopting it at a time when obtaining the proofs needed would be much harder due to the increasingly rigorous nature of the rules governing clinical trials. Finally, and probably most importantly of all, the regulatory environment has changed drastically. The first MRI guidelines of all (by NRPB in 1979) were largely established through the assurances of those in the field that they thought that everything would be all right. For example, the B0 field level of 2.5 T (as it was in the original NRPB guidelines) was set because no one consulted could think of anything to suggest it should not be that high. On the other hand, though it was known that George Radda was aiming to get a 1.9-T whole body system for MRS, no one could think of any reason why higher fields than that would be needed or be economic if they actually turned out to be practical. Things like the 5-gauss line were the result of completely anecdotal results. MRI was the first major equipment modality to be reviewed by the FDA (as a result of an Act of Congress in 1976). The applications for approval made by Technicare (then a Johnson & Johnson subsidiary), Diasonics, and Picker were reviewed at the same panel meeting. The panel concluded that none of the applications was approvable by itself, but that they could approve all three as a group. Technicare had plenty of animal data, while the other two companies had clinical data from their collaborators at UCSF and Hammersmith, respectively. Modern attitudes to competition and fears about litigation would stop such a generous process in its tracks. Then again, imagine the problems we would experience nowadays if we had to go into the regulatory process with a totally strange and largely unproven modality admitting that it used radiofrequency powers thousands of times greater than those of mobile phones and audio frequency fields massively greater than the levels that currently cause concern about the induction of tumors in children. (The paper about the incidence of childhood cancer in those families living near power lines was published in 1980; but, while it was beginning to cause anxiety to the electricity supply industry, it had not had any impact on regulators such as NRPB, who were then primarily focused on x-rays, which are clearly of much higher energy.) The level of proof of safety needed for the introduction of MRI that would now be required would be impossible to fund, particularly if the possibility of non-thermal effects are included (as they would be), and the environment in which the technique's clinical success has allowed it a degree of generosity in each increase in guideline levels would simply not exist. Thus, the invention was supremely timely and, as millions of patients have discovered, hugely important. It is no more than justice that the Nobel Committee has rewarded Paul Lauterbur and Peter Mansfield in the way it has this year. I have drawn very heavily from three authors in writing this editorial, and cite them rather than the great many original papers. Most importantly, the late E. Raymond Andrew, who was a most intimately involved yet extremely dispassionate judge, is the source of much of the material. His chapters “Perspectives in NMR imaging” (In: Partain CL, James AE, Rollo FD, Price RR, editors. Nuclear magnetic resonance (NMR) imaging. Philadelphia: W.B. Saunders; 1983), “NMR in medicine: a historical review” (In: Partain CL, Price RR, Patton JA, Kulkarni MV, James AE, editors. Magnetic resonance imaging (MRI). 2nd edition. Vol. 1. Clinical principles. Philadelphia: W.B. Saunders; 1988), and “Magnetic resonance imaging: a historical overview” (In: Young IR, editor. Methods in biomedical magnetic resonance imaging and spectroscopy. Chichester, UK: Wiley; 2000), or in the Encyclopedia of NMR (Wiley; 1998) is a superb summary of the early period of development. These were extended versions of an earlier “Historical review” in the British Medical Bulletin (1984;40:115–119). The material about the gestation of the various modalities came from Young's “Review of modalities with a potential future in radiology” (Radiology 1994;192:307–317). Third, I have used Donald P. Hollis' book “Abusing cancer science: the truth about NMR and cancer” (Chehalis, Washington: Strawberry Fields Press; 1987; ISBN 0-942033-15-9) for a background view of the controversy involving Raymond Damadian. Finally, I have relied heavily on my own memories and prejudices, which can be blamed for any inaccuracies and inadequacies of perception. I am very grateful to Paul Bottomley who has had the patience to read this, and the wisdom and kindness to suggest many improvements. Paul has as long an experience of MRI as I have (quite possibly even longer), and his insight into what happened in the early years is extremely keen.
No takes yet. Share an insight, caveat, or question.
I. R. Young (2004) studied this question.