Translating a Dream into Reality
Translating a Dream into Reality
Birth of MRI and Genesis of a “Big Science”
Abstract and Keywords
Celebration of MRI as “the ultimate imaging technique” is today neither uncommon nor unwarranted. But, in the 1970s, scientists and nonscientists alike were unsure whether it could ever be developed. Apart from theoretical and technical difficulties, there were a variety of other issues and concerns that stood in the way of MRI's (it was not even called MRI then) emergence as a medical imaging technology. The birth of MRI, this chapter shows, was the outcome of a variety of unpredictable and contingent, albeit hierarchical, entanglements that stretched across both time and geography. It argues that distinctions between invention, development, and diffusion of technology are, in practice, messy and muddled. Different stages in the life cycle of a technology, as the history of MRI illustrates, are often folded onto each other. It also argues that, although contingent upon circumstances, the development of MRI was a hierarchical and exclusionary process. The transformation of MRI research into a “big science,” among other things, had the consequence of privileging some research groups and excluding others and as such it also had a dramatic impact on the transnational geography of MRI research and development.
MR imaging is fulfilling the promise of becoming the ultimate imaging technique.
I take this opportunity to apologize to MRI pioneers in the audience because I never believed MRI would work, like Rutherford, who said anyone who believed nuclear radioactivity would be useful “is talking moonshine.” However, I was only one of the unbelievers.
Celebration of MRI as “the ultimate imaging technique” is today neither uncommon nor unwarranted. But, in the 1970s, scientists and nonscientists alike were unsure whether it could ever be developed. Thus, when it came to MRI, even Ervin Hahn, one of the pioneers of NMR research, was an unbeliever, dismissing its medical possibilities as “talking moonshine.”1 Long after it was first proposed, NMR imaging of macro objects, specifically human bodies, remained a distant dream. Indeed, as Paul Lauterbur recalled, “MRI's death certificate was signed several times” during this period.2 Apart from theoretical and technical difficulties, there were a variety of other issues and concerns that stood in the way of MRI's emergence as a medical imaging technology.
As this chapter will show, the birth of MRI was the outcome of a variety of unpredictable and contingent, albeit hierarchical, entanglements that stretched across both time and geography.3 Contrary to Everett Rogers's widely influential “diffusion of innovation” thesis, the life cycle of MRI cannot be portrayed as an S curve.4 The issue here is not simply a matter of black boxing technical facets of technology development that diffusions models commonly do.5 Diffusion models such as Rogers's hypostatize the history of technology: they present a caricatured and, in essence, linear understanding of invention and diffusion.
(p.38) My argument in this chapter is twofold. First, I argue that distinctions between invention, development, and diffusion of technology are, in practice, messy and muddled. Different stages in the life cycle of a technology are often folded onto each other. The development of MRI was definitely not linear. Despite promising contributions in the 1970s by Paul Lauterbur, Raymond Damadian, and Peter Mansfield, who are usually credited with the invention of MRI (see chapter 1), the birth of MRI occurred much later.6
Second, I argue that, even though it was contingent on circumstances and a result of bootstrapping and bricolage of different ideas, techniques, business interests, and health-care concerns, the development of MRI was also a hierarchical and exclusionary process.7 It occurred through innovative “boundary work” that was propelled by the regular flow of epistemic and technological objects and of scientists across disciplines, institutions, and nations.8 As it proceeded, new technoscientific trails emerged and some older trails got disconnected. It also resulted in MRI research becoming a big science, which, among other things, also had the consequence of privileging some research groups and excluding others.9 This chapter analyzes the hierarchical entanglements of this shift with respect to both industry's involvement in and the transnational geography of MRI research.10 At one level, however, NMR research was already a “big science” well before the emergence of MRI (see chapter 4).11 The shift of NMR to biomedical imaging thus made the related MRI research an even bigger science.
Contingency, Bricolage, and Emergence of a “Big Science”
It is clearly thanks to Mother Nature's good graces that NMR in human subjects is possible at all. If it took, instead of seconds, hours for spins to repolarize, the technique would be impractical.
—Felix W. Wehrli, “The Origins and Future of NMR Imaging,” 1992
The initial proposals of Damadian, Lauterbur, and Mansfield definitely generated interest in an NMR imaging technology for biomedical purposes. Nevertheless, scientists soon realized that “Mother Nature's good graces” did not take them very far.12 Although, as the section epigraph suggests, the few seconds it took for protons polarized by a magnetic field to return to a relaxed state offered hope, imaging the human body was a different matter altogether.
Before proceeding further with the history of MRI development, let us briefly consider the different components of MRI (see figure 2.1). I must emphasize that these components were not known in advance: The (p.39) development of MRI, thus, was not simply a result of implementation of certain ideas and techniques. In fact, these components continue to transform even at present, offering new diagnostic and research possibilities.
A magnetic resonance image is a map of magnetic properties of biochemical compounds inside the body. Atoms of hydrogen, which are abundant in the body because of the presence of water and fat, become sources of spatial and functional data about the body in NMR imaging (theoretically, any atom that has an odd number of protons, that is, atoms that have magnetic moments, can be used). John Mallard, whose group at Aberdeen, Scotland, became one of the most important contributors to the development MRI, explained the process of NMR imaging in lucid detail:
The protons of hydrogen [atoms] spin and have an associated magnetic field.… If they are placed in a magnetic field, they will line up more or less parallel to that field and will precess around it … The rate or frequency of the precession is proportional to the magnetic field strength in which they are placed.
In nuclear magnetic resonance, one makes uses of this precession to study the atomic nuclei and their surrounding by irradiating them with electromagnetic radiation of exactly the same frequency as their precession. At that frequency they absorb energy from the radiation—a resonance absorption—and change their alignment relative to the applied magnetic field.
After a 90° pulse, the nuclei have surplus energy which they radiate to their surrounding at the same resonant frequency. From a sample containing a large number
The length of time needed for this is associated with the environment of the protons because the easier it is for them to pass energy to neighbouring atoms of the lattice of which it is a part, the quicker they can return to their original state; the strength of the signal falls exponentially with time at a rate characteristic of their environment.13
Thus NMR imaging emerged as a result of the entangled development of several technical components: (1) a homogeneous and high-strength magnet, used to make the protons of hydrogen or other elements precess and align with the magnetic field that it produces; (2) a gradient applied orthogonally to the magnetic field in order to map the position of particular protons; (3) coils placed over particular body parts being imaged to send radio-frequency pulses to the protons of the constitutive atoms of that body part and to then measure the signal when the protons of these atoms reradiate excess energy as they go back to their original position; and (4) pulse sequences (e.g., a 90°-180°-90° sequence) to make the protons of these atoms absorb and then release that energy. The data that are collected through this technological assemblage are then converted into images using a variety of software.14
Today, all this may seem a complex yet straightforward application of certain ideas or techniques. In the 1970s, the situation was dramatically different, however, as Mallard made plain: “There were times during the next six years until 1980, when [a NMR imaging technology for the human body] seemed a mistake.”15
And Mallard was certainly not alone in thinking this way about NMR imaging. When MRI eventually emerged, it did so only through multiscale entanglements that stretched across several disciplines and nations and that required complex alignments of techniques, business interests, and health-care concerns. Indeed, it was these multiscale and multidomain entanglements, along with the high economic cost (large magnets in particular were very expensive), that gave rise to the “bigness” of MRI research.
In the 1970s and the early 1980s, research groups tried out different types of magnets, coils, pulse sequences, and imaging techniques, not to mention different kinds of images.16 None of their choices was straightforward, and the development process was beset with uncertainties. Moreover, technical issues were intertwined with business and health-care concerns. Imaging of human beings, for example, required manifold enlargement of the magnets being used for NMR research. Such a shift in scale was not (p.41) only quite expensive; it also posed various technical problems and health concerns.
In order to fit a human body inside, the NMR magnet had to have a large bore. At the same time, its magnetic field had to remain homogeneous and of high strength. When scientists decided to make the shape of the magnet cylindrical in order to produce a high-strength and homogeneous magnetic field over the human body, new technosocial complications arose. Patients, who had to remain inside the resulting cylindrical tube for a long enough time (in the 1970s and early 1980s, it was far longer than the 20–30 minutes it takes today), started to feel claustrophobic, which further aggravated the problem of motion artifacts from a patient's physical and physiological motion (e.g., heartbeats).17 Consequently, various averaging (or gating) techniques were developed to allow for the impact of a patient's motion and other such effects that we rarely hear about because they commonly become invisible within the category of “noise.” Even though motion artifacts continued to be an important concern for NMR imaging, engaging them also led to new technological developments, such as MR angiography.18
There were also concerns about the heating effects of the magnetic field on human tissue.19 And the use of powerful magnets raised still other safety concerns. In one instance, the NMR magnet was reported to have pulled a large oxygen cylinder into its bore.20 There were also concerns with regard to the impact of the magnet on metallic objects implanted in the body. Hence development and deployment of MRI also necessitated extensive changes in medical practices.
It called for reorganization of the hospital space, for example. Metallic objects in and near the building posed a problem because they interfered with the magnetic field used for NMR imaging. Ron Schilling, the former president of Diasonics, one of the first MRI manufacturing companies, told me of a particular case where his people had no idea why a machine kept producing motion artifacts. But, eventually, they found out why. The hospital was near railroad tracks; whenever trains passed by on those tracks, they also transferred motion artifacts to the NMR images.
Concerns about NMR's powerful magnetic fields also necessitated development of shielding techniques. David Hoult, who worked at the National Institutes of Health in Bethesda, Maryland, on their NMR imaging project in the 1980s, recounted the issues in relation to shielding of the magnet: “Everybody was very scared of the effects of steel on [field] homogeneity … people were proposing … stand-alone buildings with no steel of any (p.42) sort in them and so on.”21 After considering the techniques that were being envisaged, he proposed one of his own:
People had already had the idea of crudely putting steel outside the magnet … for shielding purposes. The old Picker company, EMI, had tried to shim magnets with steel on the outside of them. What I realized was that if you could put steel on the inside you ought to mathematically be able to analyze how to shim them. The problem was how to do the mathematics.22
The process was not only messy, but also extremely uncertain, as Hoult explained: “We essentially developed the theory as we went on, but, it was a success story in the end.”23
Development of the magnet and its alignment within the technological assemblage of NMR was similarly entangled with yet another set of technosocial concerns, which also required bootstrapping and bricolage of different techniques, materials, and clinical and business interests. In the 1970s and the early 1980s, there were very few companies that could build magnets to the specifications required for NMR imaging. For a long time, Oxford Instruments, based in Britain, supplied magnets to many of the groups engaged in the development of MRI. Derek Shaw highlighted the complexity and uncertainty of choosing the right ones: “Working, as I was in this period [1978–83], for Oxford Instruments was in some ways like being in an Alan Ayckbourne play. Diverse characters from all the medical imaging companies would come in to discuss their own ‘secret ideas’ and specify their own unique magnet requirements, resistive/superconducting, 0.15 T/1.5 T, four coil/six coil.”24
“The typical human-sized imaging magnets [in the second half of 1970s] were four coil resistive air core units,” Lawrence Crooks, a key scientist of the University of California, San Francisco (UCSF) group, recalled.25 But the UCSF group decided instead to use a superconducting magnet for the NMR imaging machine they were developing.26 Although they were unsure whether this would work, their collaborating partners at Pfizer were quite enthusiastic about its prospects. Indeed, “one thought it sounded so sexy that no doctor would pass it up.”27
National interests were also frequently at play in these transactions. Crooks told me that Oxford Instruments first supplied a superconducting magnet to the Electrical and Musical Industry (EMI) group, based in Britain, even though UCSF had ordered it before EMI.28 Such seemingly trivial issues were important for the research groups, because these groups were in the race to produce the first images of the human body (see also chapter 1).
(p.43) The University of California, San Francisco, eventually received a 15 MHz superconducting magnet from Oxford Instruments and was able to produce images of the brain that received widespread attention. These images established UCSF as one of the key centers for MRI development.29 Nevertheless, the choice of a superconducting magnet for NMR imaging continued to be contested on both technical and economic grounds.30 Joseph H. Battocletti estimated the annual operating cost of a superconducting magnet without liquefier in 1979 to be around $75,000, as compared to $0 for a permanent magnet. He went on to argue that the “permanent magnet structure for whole body imaging is not only practical, but its initial cost is less than that of a resistive magnet, and much less than that of a super-conducting magnet.”31 For these reasons, Damadian and his group at Fonar shifted to permanent magnets, even though the image of the chest that lent legitimacy to Damadian's claim for priority in the invention of MRI had been produced on a superconducting one (see chapter 1).32
Despite their high cost, however, there was a marked shift toward superconducting magnets for NMR imaging in the 1980s, particularly in the United States. In 1983, 44 percent of the magnets being used for NMR imaging in the United States were resistive electromagnets and 49 percent were superconducting. By 1985, however, as MRI machines started to receive approval for clinical use from the FDA, these figures had changed dramatically, with only 7 percent of the machines using resistive electromagnets and 87 percent superconducting ones.33 Such a shift occurred in part because of the technical advantages of superconducting over resistive magnets in providing stable magnetic fields. It was also intimately tied to business strategies, as we shall explore in greater detail in chapter 3.
The development of imaging techniques, pulse sequences, and coils was open ended: time after time, technical possibilities that seemed theoretically promising simply did not work. Ian Young, who worked for EMI, recounted one such instance:
The Department of Health man … was a great fan of Peter Mansfield and insisted that … doing echo-planar [a fast imaging technique proposed by Mansfield] was the only thing to do [but] we just could not get signal-to-noise ratio sums to come out right.34
It may seem that the coils for sending radio frequency pulses to a particular body part in order to gather T1, T2, or proton density information from that part might have been easier to develop. But even their development was uncertain and contingent on circumstances. “In 1981, we got (p.44) absolutely superb images,” Alex Margulis recalled. “That was again luck: … we put Leon Kaufman in the machine first because Larry Crooks was running it. Kaufman was big and fat, and he filled the coil. When Kaufman was running it and Crooks was in, the images were miserable because Crooks is very thin.”35
This experience made the UCSF group realize that the imaging coil should fit tightly around the body, and, when it did, they “got probably the best images.”36
Software for image construction similarly emerged through bootstrapping a variety of techniques while grappling with the context at hand. Its development, in the first instance, was entangled with possibilities offered by the computer. Edwin Becker, who was associated with NMR research at the National Institutes of Health for fifty years, emphasized the key role of computers:
The computer really was critical. None of this could be done without special purpose computers.… Back in the late 1970s, we worked with people [at the Computer Division of the NIH] on structures of molecules, not proteins at that stage, smaller molecules.… The question was how did they fit together? And one could look at this with the computer graphics techniques.… It was very helpful to interpret the results.37
However, aligning computers to the technological assemblage of NMR was no simple matter. The hardware of the computer had to be redesigned and new software and computation techniques developed. Starting in 1964, Richard Ernst and his colleagues, initially at Varian Associates in Paolo Alto, California, and then at the Swiss Federal Institute of Technology (ETH) in Zurich, pioneered the Fourier transform NMR, which eventually became the standard technique for image construction, as opposed to the back-projection method for computed tomography (CT) scanning, used at first by Lauterbur to construct NMR images in the 1970s.38 Jim Hutchison, who belonged to the Aberdeen group in Britain, found that implementing the Fourier transform technique required significant adaptation. He and his colleagues eventually developed the “spin-warp” method of imaging. According to Bill Edelstein, a key contributor to MRI development and one of Hutchison's collaborators at Aberdeen at that time:
Our method turned out to be much better in practice. First, taking up longer times, as did the Ernst approach [the nonmodified Fourier transform method], would lose signal. Also, our method was substantially immune to field inhomogeneities [that led to artifacts in images], whereas the Ernst method was not.39
(p.45) Thus the entangled history of MRI extended across space and time. On the one hand, it involved bootstrapping existing techniques and technologies (e.g., the computer); on the other, it entailed transforming old and producing new technoscientific trails. Moreover, entangled NMR imaging research efforts across different nations were not only imbricated in existing transnational hierarchies; they also created new ones, as we shall see. For now, however, let us return to the development of the magnets, perhaps the most important and yet uncertain and openended aspect of MRI research, dramatically affecting both its culture and its transnational geography.
Apart from other uncertainties, researchers were not sure about the optimal magnetic field strength for NMR imaging. Early indications were that the magnet should be “10 MHz maximum (0.23 T) for body imaging.”40 But the group at the University of California, San Francisco, had exceeded that theoretical maximum and produced images at 15 MHz. Indeed, “the lack of RF [radio frequency] penetration problems and effective head and body coils at 15 MHz with superior [signal-to-noise ratio] began a race to ever higher magnetic field strength.”41 Even though UCSF scientists had showed that NMR images could be produced at 15 MHz (or 0.3 T), they did not believe that their new maximum could itself be exceeded. Yet it was, by a hands-on approach and through intertwining of technical and business concerns.
Paul Bottomley, a member of Raymond Andrew's group in Nottingham who had authored one of the most influential papers on the limiting effects of magnetic field strength on the penetration of radio frequency pulses joined General Electric Medical Systems' Corporate Research and Development Center in the United States in 1980. Later that year, GE also hired Bill Edelstein, who was earlier a member of Mallard's NMR imaging group in Aberdeen. These two scientists were to conduct research and development in the field of NMR spectroscopy because “GE had concluded that NMR imaging could never compete with X-ray computed tomography [CT scan] in terms of signal-to-noise ratio per unit time.”42 Indeed, it was not until early 1982 that GE seriously thought of investing in NMR imaging.43
The turning point came in December 1981 with the exhibition of NMR images by Diasonics, Siemens, Philips, Picker International, Technicare, and some other MRI manufacturing companies at the annual meeting of the Radiological Society of North America (RSNA). Impressed by the quality of these exhibits, GE decided to enter the field of NMR imaging in the belief that diagnostic MRI machines could impact its CT market.44 This change in business strategy had a direct impact on the technical choices.
(p.46) GE decided to opt for a magnet of lower field strength because, in the early 1980s, scientists believed that NMR imaging could not be conducted at high magnetic fields. But since GE had already placed an order for a high-field magnet (1.5–2 T; 5–10 times higher than the ones being used for NMR imaging at that time) in their effort to focus on NMR spectroscopy, the company had to reorient its strategy and respond to the changed context. The plan, as Bottomley recounted, “was to obtain a few spectra at high field when the magnet arrived, then turn it down to 0.15 T.”45 In the process, however, the GE researchers ended up devising the “birdcage” head-imaging coil and found that images of the head could be produced even at 1.5 T with that coil.46
Again, hands-on work and bootstrapping of technical and social concerns led to extending the accepted theoretical boundaries.47 GE's high-field NMR provided a much higher signal-to-noise ratio and hence higher-resolution images. But because the magnet accounted for nearly half the cost of an MRI machine and required alignment of a range of interests and expertise, the shift to high-field NMR imaging also made the related MRI research an even bigger science. The transnational geography of MRI would change dramatically in the early 1980s, particularly in relation to Britain and the United States.
Shifting Geography of NMR Imaging Research
Although NMR imaging was initially proposed in the United States, much of the early development was carried out by small groups of physicists and engineers in Britain.
Despite the daunting uncertainties and high cost of the new technology, research groups and multinational companies from several nations joined the race to develop MRI in the late 1970s. Britain and the United States were seen as the main centers, however. Research in these two countries, even though entangled, was also marked by competition.
“Billed as a back-to-back showdown between the British and the U.S. groups,”48 the 1981 International Symposium on Nuclear Magnetic Resonance Imaging, held at Winston-Salem, North Carolina, in fact marked the one-sided dominance of British scientists. In his notes on the Symposium published in the Journal of Computer Assisted Tomography, William Oldendorf not only reported that “of the 31 invited speakers, 14 had done their work in (p.47) the United Kingdom,” but all the NMR imaging developments he discussed were the work of research groups there.49 Oldendorf went on to say:
This lopsided national representation was appropriate because most of the work on NMR imaging has been from the U.K., although the key quantum jump that started the field was by Paul Lauterbur.… This shows the inability of the U.S. technology to follow up on a good lead.… My personal bias would answer this by pointing out that half of the U.S. research and development effort is put into defense research.50
Not surprisingly, therefore, MRI development has often been portrayed as a British achievement. In November 1978, New Scientist published the image of Dr. Hugh Clow's skull in an article titled “Britain's Brains Produce First NMR Scans.”51 Yet another journal article proudly claimed, “Magnetic Resonance Imaging: Another Scottish First.”52 On its “Medical Imaging Timeline” website, the Engineering and Physical Sciences Research Council of the United Kingdom describes the British achievements in NMR imaging thus: “Peter Mansfield and his team at the University of Nottingham pioneered the use of MRI in medicine. John Mallard's team at Aberdeen University developed the ‘spin-warp’ technique that produced MR images quickly in the third dimension.”53
The dominant role of British scientists notwithstanding, NMR imaging research in Britain started in transnational contexts. Mansfield, who until early 1973 was engaged in mapping the internal structure of solids with NMR, first heard about the possibility of biomedical imaging during the First Specialized Colloque Ampere in Krakow (see chapter 1). Raymond Andrew, who headed another group at Nottingham, got involved with NMR imaging when he and his colleagues “heard Paul Lauterbur talk about NMR imaging” during the International Society of Magnetic Resonance (ISMAR) meeting in Bombay (present-day Mumbai) in 1974. The Aberdeen group, because of Mallard's involvement with medical engineering projects such as electronic spin resonance (ESR), was aware of both Damadian's and Lauterbur's initial work with NMR imaging in the United States.54 EMI, on the other hand, “began work on magnetic resonance imaging in 1974, simply because the company wanted to be involved in any, and all, modalities with any potential to rival its then burgeoning CT X-ray business.”55
The enthusiasm and effectiveness with which the British groups took up NMR imaging were unparalleled. Andrew recounted his group's response on its return from the ISMAR conference in Bombay: “Waldo [Hinshaw] and Bill [Moore] soon devised an alternative approach and on our return to Nottingham, Waldo, working in my laboratory with our Bruker spectrometer, was producing NMR images in a few weeks.”56
(p.48) Mallard described his Aberdeen group's initiative in quickly putting together an NMR imaging project in similar terms: “Since we were already to the fore in computed tomography and had all the computer programs to hand, we were very quickly able to explore the possibility of NMR imaging. We obtained the first image of a whole mouse in Aberdeen in March 1974.”57
By 1974–75, British groups were already ahead of their U.S. counterparts. In part, this happened because Britain was uniquely placed in relation to interdisciplinary research. It had a long history of interdisciplinary collaborations, particularly between physicists, engineers, and practitioners of medicine.58
The Roentgen Society, founded in 1897 with Sylvanus Thompson, a physicist and electrical engineer, as its first president, was an important precursor of such interdisciplinary engagements.59 Jeff Hughes recounted their important role in his introduction to the Wellcome Witness Seminar “Development of Physics Applied to Medicine in the UK, 1945–1990”: “The emergence of medical physicists and the increase in their numbers—from a handful of pioneers at a few leading hospitals in the 1910s to early 1930s to practitioners all over the UK in the later 1930s—marked the birth of a new profession.”60
Eventually, in 1943, Britain's Hospital Physicists' Association (HPA) became “the first national body … in the world” that actively promoted medical physics research.61 HPA was not only fifteen years ahead of a similar national body in the United States, the American Association of Physicists in Medicine (AAPM); it was also a motivating and guiding force for the AAPM.62 Hence, even though John Mallard was the only trained medical physicist to head a British NMR imaging group, the history of engagement between physicists, engineers, and medical practitioners provided a ready-made “trading zone” for interdisciplinary research in Britain.63
The Nottingham groups, unlike the Aberdeen group and the groups associated with EMI, did not have strong collaborative links with clinicians in the early phase of MRI development, even though Raymond Andrew had played a pivotal role in establishing a medical physics group at Nottingham.64 Nevertheless, as Peter Mansfield recounted, his group did work “with medical colleagues in Nottingham.”65 He also found the “support and comments” of Donald Longmore, a cardio surgeon, who was engaged in NMR imaging from the beginning, “extremely gratifying in those early days.”66
This history of interdisciplinary research must have played a significant, albeit unrecognized, role in the decisions of the Medical Research Council (p.49) (MRC) and the Department of Health to fund several projects in Britain, particularly in the 1970s, when there was little support for development of MRI elsewhere, including in the United States.67 In the “early years of the 1980s,” as Gordon Higson recalled, “the Department [of Health] was paying more than a million pounds, maybe up to a peak of about a million and a half pounds a year, into the various NMR activities, and this was out of an R&D budget of about £4 million.”68
In the United States, apart from Damadian's team, which operated on a much smaller scale, the UCSF group was also able to create a productive “trading zone” between physics, engineering and medical science in the later 1970s. This cross-disciplinary collaboration for NMR imaging at UCSF, Alex Margulis explained, had genealogical links to a similar engagement with CT scanning development.69 Such cross-disciplinary NMR imaging groups were rare in the United States, however, and also not as successful as the British ones.
Interestingly, at the time of their greatest success in the 1970s and the early 1980s, the British groups were also very international in character. All of these groups (at Nottingham, Aberdeen, and EMI) had key scientists who were not of British origin. For example, William (Bill) Edelstein of the Aberdeen group was an American. Waldo Hinshaw and P. A. Bottomley, who belonged to Raymond Andrew's group in Nottingham, were from the United States and Australia, respectively; Graeme Bydder, a central figure in EMI's NMR imaging project, was from New Zealand. The importance of this international presence can be gauged from the size of NMR imaging research groups in the 1970s, which usually had only three to five key scientists.
The autobiographical essays of several NMR scientists in The Encyclopedia of Nuclear Magnetic Resonance, volume 1, further underscore the transnational character of NMR imaging research in Britain in the 1970s.70 If there were hardly any scientists from India at that time, we should keep in mind the effect of hierarchical restrictions (e.g., visas) on the mobility of scientists from India and the impact of colonial rule.71 As the world moves toward more “flexible citizenships,” many more scientists from countries such as India are working in other parts of the world, leading to novel technological developments (see chapter 4).72
The broader point here is that if a particular geographical region is or has been a center of technoscientific research, it is because it has been a direct or indirect destination of transnational flows of people, knowledges, and artifacts.73 Perhaps the strongest evidence for such a thesis with regard to NMR imaging is the shift that occurred in Britain in the early 1980s. In the (p.50) 1970s scientists went there even when they had jobs in the United States, as Waldo Hinshaw's case illustrates:
In late 1974, I accepted a position with Professor Irving Lowe at the University of Pittsburg. During my stay there, Professor Andrew and Bill Moore applied for and were awarded a grant from the Medical Research Council to develop an NMR imaging system. So I returned to Nottingham and, for the next couple of years, worked flat-out to put together an imaging system for 3-in. diameter samples.74
By 1983–84, however, nearly two years after William Oldendorf had claimed that the Americans were unable to compete with the British because so many American scientists were involved in defense-related research, the transnational geography of NMR imaging research had changed dramatically. Scientists from Britain and other parts of the world had moved to the United States, which now became the primary site for MRI development. During this period, as John Mallard ruefully recalled, the culture of NMR imaging research also changed. Exclusively university-based groups could no longer compete: “Due to the much greater financial and human resources that the major multinationals could bring to bear, university teams in research laboratories were gradually pushed out of further development of NMR imaging.”75
A significant reason for this shift was the transformation of MRI research into an even bigger science, which, in turn, was entangled with the United States becoming the largest market for MRI machines (see chapter 3). Consequently, British groups found they could no longer compete. In 1988, Thomas Redpath, who had joined the Aberdeen group as a Ph.D. student, wrote to Gordon Brown, former prime minister of Britain and then a member of Parliament, explaining the plight of NMR imaging research in Aberdeen and, more broadly, in Britain:
The Government response has been to cut Government and National Health Service funding so that morale at departments like my old one at Aberdeen have taken a battering.… Underfunding of the Medical Research Council (MRC), has made it extremely difficult to fund pioneering work of the kind that led to the development of MRI.… Most of colleagues from the MRI group in Aberdeen have emigrated—4 to the USA, 1 to Germany and 1 about to leave for Norway.76
By the time Redpath wrote to Brown, however, the transnational landscape of MRI research had already changed decisively, and Britain was no longer the center. The mid-1980s also marked another shift: the MRI industry was radically transformed and its transnational geography changed as well.
With 35,000 design engineers, and revenues exceeding New York State's annual budget, business is booming at General Electric—but there is more to being best than “big.”
Studies of the role of industry or particular firms in relation to technological innovations, such as David Bak's analysis of GE quoted above, commonly reflect two kinds of elisions.78 On the one hand, they overlook contingent and emergent aspects of technoscientific innovations or subordinate these to the ability of firms to adapt and innovate.79 On the other, they explain away the hierarchical and exclusionary effects of firms' practices as exceptions or market distortions (e.g., as monopolistic practices).
“Teamwork, shared ideas, targeted research, and technical alliances” were, of course, crucial for GE to “face the challenges of world competition” with regard to MRI development.80 Yet, to constitute the functioning of GE only through such attributes is to hypostatize not only the technoscientific innovation process, but also our conceptions of ingenuity and competition. GE, as we saw earlier, did not even believe that NMR imaging was possible until as late as December 1981.
EMI was the first in the medical equipment industry to start research on NMR imaging.81 Not long after it did so, in 1975–76, Leon Kaufman and Alex Margulis in collaboration with Jay Singer (from the University of California, Berkeley) and Pfizer brought together a group of scientists at the University of California, San Francisco, for the development of MRI.82 Another NMR imaging company established during this period was Fonar, incorporated in 1978 with capital Raymond Damadian had raised through private contributions.83 By 1976–77, several multinational companies, namely, Philips Medical Systems, Bruker Instruments, and Siemens Medical Systems, had also joined the NMR imaging fray.84
The involvement of the industry was, in part, propelled by some significant breakthroughs in NMR imaging. In 1977, Waldo Hinshaw, who was working with Raymond Andrew, “published an in vivo image through the wrist in Nature and [Peter] Mansfield and [A. A.] Maudsley published the image of a finger in the British Journal of Radiology.”85 The image of the chest that Raymond Damadian and his group produced also attracted a great deal of attention (see chapter 1). Even though the Aberdeen scientists had not been as much in the spotlight, the spin-warp method that they developed was undoubtedly one of the most important contributions to (p.52) the development of MRI (GE's high-field imaging used this technique, for example). The Aberdeen group, which by 1981 had conducted clinical trials on nearly 900 patients on its NMR machine, the Mark I, also provided a significant corpus of NMR imaging studies (see figures 2.2 and 2.3).86 In the United States, the images of the brain produced by the UCSF group in the early 1980s were another important contribution to the development of MRI (see figure 2.4).87
These achievements, which occurred parallel to the development of the technological assemblage described earlier, illustrated that in vivo NMR imaging of the human body was possible, although the images were still
Toward the end of the 1970s and in the early 1980s, when the culture of MRI research was changing, the “relative position of firms changed dramatically” as well.89 The shift was most evident in Britain, where EMI, the frontrunner in the NMR imaging industry, saw the fortunes of its CT scanning, NMR imaging, and music divisions rapidly decline.90 EMI's CT market share in the United States, for example, dropped from 100 percent in 1973 to 41 percent in 1977, with the company closing out its CT manufacturing completely in 1980.91 EMI's NMR imaging division suffered an even worse fate and was sold to the General Electric Company (GEC) of Britain in 1981.
Ian Young informed me that EMI had first decided to sell its NMR imaging division to GE (U.S.) on 31 December 1981, but Lord Arnold Winestock of GEC pulled off a last-minute coup. Meeting with the British home secretary, Winestock asked that the EMI division be sold instead to GEC in the national interest; EMI agreed.92 GEC thus became an important MRI development company. After it later acquired Picker, a CT manufacturing company based in the United States, it became Picker International and moved (p.54)
Another British MRI company, M & D Technology, established by John Mallard in 1982 with £1.4 million capital raised mostly in the city of Aberdeen, had a much shorter life. With a “blue skies” grant from Britain's Medical Research Council (MRC), the Aberdeen group had developed its hugely successful whole-body NMR imaging machine, the Mark I.94 It could not, however, find funds to upgrade this machine until Asahi Chemical, a Japanese company, gave them £283,000 for this purpose. In 1982, with the capital generated from Aberdeen, M & D Technology sought to manufacture and supply its next-generation machine, the Mark II. Three years later, however, (p.55) in 1985, the investors withdrew their support, and M & D Technology was “taken over for a song by a U.S. firm, Basic American Medical.”95
Thus, in the second half of the 1980s, not only did British scientists move to academic institutions and companies in the United States; British scientists and industrialists also chose to set up manufacturing bases in the United States. Just as Picker International had moved its operations to Cleveland, Ian Pykett and Richard Rzedzian of Mansfield's Nottingham group moved to Boston to set up Advanced NMR Systems for the manufacture and marketing of echo-planar techniques for MRI.96
The fortunes of U.S.-based MRI companies changed as well, though, in contrast to Britain, this did not lead to a flight of manufacturing from the United States. Pfizer, for example, sold off its “rights to all patentable NMR technology developed under the UCSF-Pfizer agreement” to Diasonics in 1981.97 Engaged at the time in the development and manufacture of ultrasound technology, Diasonics seized the opportunity to provide strong support to the UCSF group and enter the MRI industry, where it became an important stakeholder.
In the late 1970s, Technicare, a wholly owned subsidiary of Johnson & Johnson, was another U.S. company to enter the NMR imaging development business. Technicare initially wanted to buy EMI's NMR imaging division and to acquire Pfizer's stake in the UCSF group.98 When neither of these prospects panned out, however, Technicare aggressively pursued its own development program. By 1983, it had the highest number of clinical placements of NMR imagers in the United States (36, or 39% of market share). But then, in the aftermath of the 1982 Tylenol crisis, Johnson & Johnson moved out of the medical imaging business, selling Technicare to General Electric Medical Systems in 1986.99 This shift, combined with aggressive marketing of its high-field MRI machines, would make GE the market leader, even though, until 1983, it had only four clinical placements of its machines, three of which were in the United States.100
These transformations in the MRI industry, though at one level contingent on circumstances, were also entangled with the shifting culture of MRI research. By the mid-1980s, one firm (or group) had little technological advantage over the other, or, as Larry Crooks observed: “Everybody [was] using everybody else's technique in some sort of mix.”101 The high-field imaging that GE had achieved was also eventually brought into the mix by other groups. Although the sharp increase in magnetic field strength made MRI research and development an even bigger science, privileging large multinational companies such as GE, Siemens, and Philips, this shift did (p.56) not settle the debate over ideal magnetic field strength for NMR imaging. Even as it emphasized the advantages of high field for NMR imaging (and spectroscopy), the GE group was also clearly aware of its drawbacks:
Balancing the above advantages of high-field operation are several potential limitations: (1) the effects of eddy currents in the body, (2) chemical-shift artifacts, (3) difficulties in the design of radiofrequency coils, and (4) radiofrequency power levels.102
Nevertheless, GE and other large multinational companies such as Siemens and Philips aggressively marketed the high-field MRI. Several of the radiologists and scientists I interviewed in 2008 remembered the lavishly illustrated brochures GE distributed during the annual meeting of the RSNA in 1982, asking them to put their buying plans on hold for GE's new high-field machines. GE advertisements in the radiology journals during this period boldly informed readers how they might “avoid being left behind as MR technology evolves,” and advised them to “make high field a high priority” and “consider what GE commitment meant to our CT customers as that technology matured. Then give your General Electric representative a call.”103
The fate of the NMR imaging firms thus seems to show a striking similarity with what has been described as the trajectory of the CT scanners. “As the dust settled,” Manuel Trajtenberg argues, “it became clear that the firms that prevailed in this very competitive market were, almost exclusively, those that had from the start a deep involvement in the market for conventional X-rays. The main reason seems to reside in marketing complementarities rather than in any sort of strict technological advantage.”104 Although one could argue, as Trajtenberg suggests, that “marketing complementarities” eventually led to the success of companies such as GE, Picker International and Diasonics (that failed to succeed) had marketing complementarities of different medical imaging technologies as well, and so did EMI.
Thus the shifts in NMR imaging industry were in part contingent on circumstances. But they were also a result of the “muscle” of large multinational companies. GE's references to CT scanners in its advertisements, for example, apart from touting the firm's broader complementarities, were also a reminder of GE's market strength, which drove smaller companies such as Diasonics out of the running (Diasonics was sold to Toshiba in 1991). Ron Schilling, GE's vice president in charge of its international marketing in the 1970s before taking over as the president of Diasonics, explained GE's business strategy. Even when GE was not convinced that NMR imaging would succeed, its strategy was to make researchers follow the trajectory it wanted (p.57) to undertake.105 And the market muscle of multinational companies such as GE made it virtually impossible for others to productively and profitably follow alternative technoscientific trails.
As chapter 3 will show, the shift in the NMR imaging industry and, more broadly, the transformation of NMR imaging research into a big science were also entangled with the emergence of the United States as the largest market for medical imaging machines, which, among other things, also resulted in the renaming of NMR imaging.
(2) . Interview with Paul Lauterbur, 22 September 2000.
(3) . Bruno Latour argues that technologies embody multiple temporalities and histories because the constitutive elements of any technology belong to different historical periods. See Latour, We Have Never Been Modern, 72–76.
(4) . Rogers argues that the process of technoscientific innovations and their diffusion assumes an S shape because it goes through five stages, namely, those of (1) innovators, (2) early adopters, (3) early majority, (4) late majority, and (5) laggards. Everett Rogers, Diffusion of Innovations (New York: Free Press, 1995), 22. Although such characterization is seemingly nonlinear (an S curve), it remains parasitic to a linear construction of history of technology (invention ⊠ development ⊠ diffusion). For a critique of diffusion models of science see Latour, Science in Action.
(5) . Innovation theorists are becoming increasingly concerned about the implications of black boxing technical facets in the analysis of technological development. See Jan Fagerberg, “Innovation: A Guide to the Literature,” in The Oxford Handbook of Innovation, ed. Jan Fagerberg, David Mowery, and Richard Nelson (New York: Oxford University Press, 2005), 1–26. Within science and technology studies, the “social construction of technology” (SCOT) was among the first to provide an empirical approach to open the black box of technological innovations by analyzing them through a “multidirectional” model. See Trevor Pinch and Wiebe Bijker, “The Social Construction of Facts and Artifacts: Or How the Sociology of Science and the Sociology of Technology Might Benefit Each Other,” in Bijker, Hughes, and Pinch, The Social Construction of Technological Systems, 17–50; see also Pinch, “Opening Black Boxes.” Michel Callon extended the empirical approach for the study of technological (p.139) developments by making the role of technical, social, economic, and other factors symmetrical and contingent on circumstances. Michel Callon, “The Sociology of an Actor-Network: The Case of the Electric Vehicle,” in Mapping the Dynamics of Science and Technology, ed. Michel Callon, John Law, and Arie Rip (London: Macmillan, 1986), 19–34, and “Society in the Making: The Study of Technology as a Tool for Sociological Analysis,” in Bijker, Hughes, and Pinch, The Social Construction of Technological Systems, 83–103.
(6) . A particular consequence of framing the emergence of MRI, or for that matter any technology, in linear terms is that accounts of its invention and diffusion proliferate at the expense of attention to any other part of its history. Blume's analysis and, more recently, Joyce's study are rare exceptions. See Blume, Insight and Industry; Joyce, Magnetic Appeal.
(7) . Andrew Pickering uses the concept of interactive stabilization to argue that technoscientific practice is open ended and emergent and occurs through “accommodations” and “resistances” of human and material actors. Pickering, The Mangle of Practice, 14–20. It should be noted, though, that temporal emergence and the constitution of hierarchy and exclusion are often intertwined.
(9) . Different facets of “big science” have been analyzed, but its exclusionary impact has been rarely investigated. Sharon Traweek highlights the colonialist aspect of big science. Traweek, “Big Science and Colonialist Discourse.”
(10) . John Krige shows how American hegemony and practices of basic science in postwar Europe were coproduced. See John Krige, American Hegemony and the Postwar Reconstruction of Science in Europe (Cambridge, MA: MIT Press, 2008).
(11) . Bruce Hevly, following Derek J. de Solla Price, points out that “each generation defines big science in comparison to what went before.” Bruce Hevly, “Reflections on Big Science and Big History,” in Galison and Hevly, Big Science, 356. Because technoscientific contexts vary significantly across nations (see chapter 4), it is also important to analyze big science in terms of its geography as well as its chronology. See Traweek, “Big Science and Colonialist Discourse.”
(14) . An important advantage of NMR imaging (as compared to, for example, CT scanning) is that it could utilize a variety of parameters to collect data from inside the body and use them to construct different kinds of images (e.g., T1, T2, or proton density).
(16) . For a detailed discussion on different techniques that were being used in the 1970s and the early 1980s see Joseph Battocletti, “NMR Proton Imaging,” CRC Critical Reviews in Biomedical Engineering 11, no. 4 (1984): 313–361.
(17) . See Kieran Murphy and James Brunberg, “Adult Claustrophobia, Anxiety and Sedation in MRI,” Magnetic Resonance Imaging 15, no. 1 (1997): 51–54; H. K. McIsaac et al., “Claustrophobia and the Magnetic Resonance Imaging Procedure,” Journal of Behavioral Medicine 21, no. 3 (1998): 255–268; I. Eshed et al., “Claustrophobia and Premature Termination of Magnetic Resonance Imaging Examinations,” Journal of Magnetic Resonance Imaging 26, no. 2 (2007): 401–404.
(19) . In all such cases, “technical” choices have been inseparable from the “social” ones. Hence, for example, in his review of the heating effects of radio frequency pulses, Frank Shellock writes, “Using fast-spin-echo (FSE) and magnetization transfer contrast (MRC) pulse sequences on high-field-strength MR systems may require levels of RF energy that easily exceed” the safety limits imposed by the FDA at present. Frank G. Shellock, “Radiofrequency Energy-Induced Heating during MR Procedures: A Review,” Journal of Magnetic Resonance Imaging 12, no. 1 (2000): 34.
(21) . Interview with David Hoult, 24 January 2008.
(24) . Derek Shaw, “From 5-mm Tubes to Man: The Objects Studied by NMR Continue to Grow,” in Grant and Harris, The Encyclopedia of Nuclear Magnetic Resonance, 1:623. “T” stands for tesla(s), the unit of measurement for magnetic field strength.
(26) . The shift to superconducting magnets included Oxford Instruments, too, although, as Crooks recounts, Shaw was at first “much more conservative about superconductive magnets since Oxford was working on four coil resistive air core designs.” Ibid., 1:270.
(28) . Interview with Lawrence Crooks, 23 May 2008.
(p.141) (30) . Several studies were conducted to estimate the overall annual running costs of MR imaging in the United States. See, for example, William Bradley, William Opel, and John Kassabian, “Magnetic Resonance Installation: Siting and Economic Considerations,” Radiology 151 (1984): 719–721. The overall running cost of MRI in the United States was estimated to be nearly $1.4 million in 1990. See Ronald G. Evens and Ronald G. Evens, Jr., “Analysis of Economics and Use of MR Imaging Units in the United States in 1990,” American Journal of Roentgenology 157 (1991): 603–607.
(34) . Interview with Ian Young, 24 April 2008.
(35) . Interviews with Alexander Margulis conducted by Nancy Rockafellar, 19 and 25 March 1996, UCSF Oral History Archives.
(37) . Interview with Edwin Becker conducted by Claudia Wassamann, 15 July 2005, Office of NIH History.
(38) . Richard Ernst received the Nobel Prize for his work on Fourier Transform NMR in 1991. He licensed the resulting patent to Varian Instruments, an NMR spectrometer manufacturing company, based in the United States. Richard Ernst, “The Success Story of Fourier Transformation in NMR,” in Grant and Harris, The Encyclopedia of Nuclear Magnetic Resonance, 1:293–306. See also A. Kumar, D. Welti, and R. R. Ernst, “NMR Fourier Zeugmatography,” Journal of Magnetic Resonance 18 (1975): 69–83.
(40) . Crooks, “Field Strength Selection for MR Imaging,” 269–270. Two important studies that indicated the limits of magnetic field strength that could be used in imaging were by Hoult and Lauterbur and by Bottomley and Andrew. See David Hoult and Paul Lauterbur, “The Sensitivity of the Zeugmatographic Experiment Involving Human Samples,” Journal of Magnetic Resonance 34 (1979): 425–433; and P. A. Bottomley and E. R. Andrew, “RF Magnetic Field Penetration, Phase Shift and Power Dissipation in Biological Tissue: Implications for NMR Imaging,” Physics in Medicine and Biology 23 (1978): 630–643. The argument presented in the former article was that signal-to-noise ratio would decrease as the field strength is increased, while the latter argued that field strength increased the penetration of radio frequency pulses that are used to spatially measure relaxation times or proton density; hence, the ideal field strength was between 10 and 30 MHz.
(p.142) (41) . Crooks, “Field Strength Selection for MR Imaging,” 270.
(43) . Bill Edelstein informed me that even though GE management wanted to focus on spectroscopy, the scientists working at GE wanted to do NMR imaging. Edelstein is presently writing an article on the role of GE in the development of MRI. Personal communication with Bill Edelstein, 9 December 2012.
(46) . GE wanted to showcase these images for the annual RSNA meeting in 1982. Their research group was working hard toward producing better resolution images. But their machine broke down. Nonetheless, the high-field images produced on GE machines created a sensation during the 1982 RSNA meeting.
(47) . Even though resolution of MR images is directly correlated to the strength of the applied magnetic field, high-field imaging also posed several technical concerns. See H. R. Hart et al., “Nuclear Magnetic Resonance Imaging: Contrast-to-Noise Ratio as a Function of Strength of Magnetic Field,” American Journal of Roentgenology 141, no. 6 (1983): 1199.
(53) . Francis Smith. Medical Imaging Timeline, 21 December 2006, http://www.epsrc.ac.uk /newsevents/casestudies/2006/Pages/medicalimagingtimeline.aspx, accessed 11 August 2013.
(55) . Ian Young, “EMI's Venture into NMR: An Industrial Saga,” in Grant and Harris, The Encyclopedia of Nuclear Magnetic Resonance, 1:724. Godfrey Hounsfield puts forth a similar reason for EMI's involvement in MRI's development in the lecture he gave after receiving the Nobel Prize for his work on CT scan. See Godfrey Hounsfield, “Computed Medical Imaging,” Science 210 (1980): 22–28.
(57) . Mallard, “The Evolution of Medical Imaging,” 358. See also J. M. S. Hutchison, J.R. Mallard, and G. C. Goll, “In-Vivo Imaging of Body Structures Using Proton Resonance,” paper presented at the Eighteenth Ampere Conference, Nottingham, 1974.
(58) . Blume argues that the research groups with medical background had a different focus with regard to NMR imaging than those that did not. He writes, “For both Mallard and Damadian, by contrast, medical goals were paramount—not only rooted in experience but also sustained by the environments in which they worked. Neither had any interest in the gradual and reflexive scaling up of the apparatus that occupied Andrew, Hinshaw, and Mansfield.” Blume, Insight and Industry, 202.
(59) . Although the term medical physics was first deployed by Neil Arnott in 1828 and later by Adolf Fick in 1856, active collaborations between physicists and practitioners of medicine became common only with the emergence of X-rays. See J. S. Laughlin and P. N. Goodwin, “History of the AAPM,” Medical Physics 25, no. 7 (1998): 1235–1237.
(61) . Alan Jennings, as quoted in Christie and Tansey, Development of Physics Applied to Medicine, 28:19.
(63) . Galison uses the concept of “trading zone” to signify the modalities of interdisciplinary interactions in technoscientific practice. According to Galison, a trading zone is “an arena in which radically different activities could be locally, but not globally, coordinated.” Galison, “Computer Simulations,” 119.
(64) . As a result, Andrew's group was well placed to engage in the development of NMR imaging. See Edward Raymond Andrew, “After-Dinner Speech: Nottingham NMR Recollections,” Magnetic Resonance Materials in Physics, Biology and Medicine 2, no. 3 (1994): 143–146.
(67) . “The Medical Research Council (MRC) was established in the UK in 1920 as the successor body to the Medical Research Committee, founded in 1912.” Christie and Tansey, Making the Human Body Transparent, 3n4.
(p.144) (69) . Interview with Margulis by Rockafellar, 25 March 1996, I would like to thank Brian Dolan of the UCSF Oral History Archives not only for allowing me access to this interview, but also for providing valuable information and suggestions in relation to magnetic resonance imaging efforts at the UCSF.
(71) . Daniel Headrick put its bluntly: “Every time a new process or piece of equipment was introduced into a colony, it came with European experts to set it up and to operate it, and sometimes to pass their jobs on to their sons.” Headrick, The Tentacles of Progress, 382.
(72) . Aihwa Ong uses the concept of flexible citizenship to describe the transnational flows that characterize present-day globalization. According to her, “in the era of globalization, individuals as well as governments develop a flexible notion of citizenship and sovereignty as strategies to accumulate capital and power” Aihwa Ong, Flexible Citizenship: The Cultural Logics of Transationality (Durham: Duke University Press, 1999), 6.
(73) . Irfan Habib in his critique of essentialized arguments about Islamic sciences argues that the context in which science prospered in the Arab world was cosmopolitan and hence to classify it parochially will not be correct. Irfan Habib, “Viability of Islamic Science: Some Insights from 19th Century India,” Economic and Political Weekly 39, no. 23 (2004): 2351–2355.
(76) . Thomas Redpath to Gordon Brown, 27 May 1988. I would like to thank Thomas Redpath for providing me with this letter and other valuable information about MRI research in Aberdeen in particular.
(79) . Economists have perhaps most closely analyzed social imbrications of invention and innovation. See Joseph Schumpeter, “Entrepreneurship, Style and Vision,” in The Theory of Economic Development, ed. Jürgen Backhaus (Dordrecht, The Netherlands: Kluver Academic Publishers, 2003), 61–116, and Essays on Entrepreneurs, Innovations, Business Cycles, and the Evolution of Capitalism, ed. Richard Clemence (New Brunswick, NJ: Transaction, 1989). Nevertheless, as has been admitted by more recent analysts of invention and innovation, traditionally within economics “the innovation process itself has been more or less treated as a ‘black box.’” Jan Fagerberg, “Innovation,” 3.
(p.145) (81) . See Steinberg and Cohen, Health Technology Case Study 27, 44–45.
(82) . Alexander Margulis, “How NMR Was Started at the University of California, San Francisco (UCSF),” in Grant and Harris, The Encyclopedia of Nuclear Magnetic Resonance, 1: 484–485. Also see interviews with Alexander Margulis, 19 and 25 March 1996, and Lawrence Crooks, 31 March 1998, conducted by Nancy Rockafellar, UCSF Oral History Archives.
(84) . For a timeline of the industry's involvement with NMR imaging, see Steinberg and Cohen, Health Technology Case Study 27, 39–54.
(87) . M. Brant-Zawadzki, P. L. Davis, et al., “NMR Demonstration of Cerebral Abnormalities: Comparison with CT,” American Journal of Roentgenology 140, no. 5 (1983): 847–854; M. Brant-Zawadzki, D. R. Enzmann, et al., “NMR Imaging of Experimental Brain Abscess: Comparison with CT,” American Journal of Neuroradiology 4, no. 3 (1983): 250–253.
(88) . By the early 1980s, nineteen companies were engaged in the industrial development of NMR imagers. See Steinberg and Cohen, Health Technology Case Study 27, 45.
(89) . Manuel Trajtenberg, Economic Analysis of Product Innovation: The Case of CT Scanners (Cambridge, MA: Harvard University Press, 1990), 51.
(90) . EMI became Thorn EMI in 1979 after it merged with Thorn Electrical Industries.
(91) . Trajtenberg, Economic Analysis of Product Innovation; Will Mitchell, “Medical Diagnostic Imaging Manufacturers,” in Organizations in Industry: Strategy, Structure and Selection, ed. Glenn Carroll and Michael Hannan (New York: Oxford University Press, 1995), 244–272.
(93) . Changes in the fortunes of firms need not necessarily follow the logic of the market that the economists of innovation commonly highlight. Donald Longmore informed me that Lord (Arnold) Winestock was building the company with the idea that his son would eventually take over. But Winestock's son was detected with cancer and died, which made him lose interest in the affairs of the company and eventually led to company's downfall. Interview with Donald Longmore, 13 August 2007.
(94) . Interview with Jim Hutchison, 21 April 2008.
(101) . Interview with Lawrence Crooks conducted by Nancy Rockafellar, 31 March 1998, UCSF Oral History Archives.
(102) . P. A. Bottomley et al., “NMR Imaging/Spectroscopy System to Study Both Anatomy and Metabolism,” Lancet 322, no. 8344 (1983): 1199.
(103) . Such advertisements were very common in most radiology journals of the time. See, e.g., GE advertisement: “When physicians refer to clinically superior MR, they're referring to Signa” American Journal of Roentgenology 148 (1987): 1.
(105) . Schilling became the president of Toshiba's U.S. MRI research and development division after Toshiba bought Diasonics. Interview with Ron Schilling, 14 February 2005.