III. MAGNETIC RESONANCE IMAGING

TECHNOLOGICAL DECOMPOSITION

Initial experiments that demonstrated the principle of magnetic resonance imaging were done on existing NMR machines. Lauterbur's first projection methods were essentially manual, so that his fundamental contribution was conceptual, not technological. Oxford Industries was able to build magnets of sufficient strength and homogeneity of field to meet both research and commercial needs. Except at high magnetic field strengths, when the dielectric properties of the human body create problems, RF transmitters and receivers generally could be built using available technology. (Note, however, that General Electric's high-strength field magnets and the associated field coil design accounted in large measure for GE's rapid rise to market dominance in the early 1980s.) Displays were not a problem, and the development of commercializable MRI machines continually drew on the state of the computer art. With the exception of a number of breakthroughs in the design of gradient coils and in techniques for spatial encoding and image reconstruction[49], the evolution of MRI was one of steady improvement in image quality made possible in part by the dramatic increases in computing power at a given price available in the marketplace. Thus the evolution of technologies intrinsic to MRI involved incremental technological improvements sparked at intervals by fundamental advances. In terms of our distinction between intrinsic and supporting technologies, MRI breaks down as follows:

NMR Imaging and Its Application to Clinical Medicine

The idea of using a magnetic gradient to introduce spatial information into signals from an NMR spectrometer, which can then be converted to an actual visual image, was Paul Lauterbur's. The idea of applying NMR to cancer detection and realizing the idea in the first commercial MRI machine were Raymond Damadian's contribution. [50]

Raymond Damadian received his M.D. degree from the Albert Einstein College of Medicine at Yeshiva University, and did his internship and residency at the Downstate Medical Center in Brooklyn, part of the State University of New York system. [51] He chose to pursue a research career rather than practice clinical medicine. He joined SUNY in 1967 with a joint appointment in Biophysics and Internal Medicine. During the years before the work that led directly to publication of his first Science article, Damadian was working on cellular composition and chemical transport. This work, focusing on cell metabolism, led Damadian to believe that there should be a way to detect cancerous cells through chemical analysis rather than by relying on direct visualization under a microscope. References in a standard chemistry textbook to Nicolaas Bloembergen's finding, 20 years earlier, that NMR relaxation times T₁ and T₂ were affected by the viscosity of a fluid led Damadian eventually to consider applying NMR spectroscopy to tissue in the hope of finding differences between cancerous and normal tissue.

At the April 1969 meeting of the Federation of American Societies in Experimental Biology, Damadian and Freeman Cope agreed to conduct NMR experiments on detecting potassium in bacteria from the Dead Sea. (Cope had been working on detecting sodium in brain tissue and wanted to measure potassium in biological tissue.) They were able to borrow time on a new, pulsed FT spectrometer made by NMR Specialties of New Kensington, PA, which enabled them to observe relaxation times directly. They were successful in detecting potassium, and this led Damadian to seek support from New York City's Health Research Council for purchase of a pulsed NMR spectrometer to explore the potential use of spectroscopy "for early non-destructive detection of internal malignancies" (letter quoted in Mattson and Simon, 1996: 646). A year later, Damadian had assembled a collection of rats bearing tumors and had again received permission to use spectrometers at NMR Specialties. The T₁ measurements Damadian made of cancerous vs. normal rat tissue were the basis for his 1971 article in Science.

Subsequent investigations by Damadian and others revealed that, although the relaxation times of signals from cancerous tissue were different from those from normal tissue, they overlapped with relaxation times from noncancerous but abnormal tissue. But Damadian was initially convinced that relaxation times could be used to detect cancer, and therefore in 1972 filed a patent claim for an "Apparatus and Method for Detecting Cancer in Tissue." The patent included the idea of using NMR to "scan" the human body to locate cancerous tissue. In early 1976, he and his graduate students began working on a prototype machine, with technical assistance from the physics department at Brookhaven National Laboratory (which put him in touch with people designing superconducting magnets for the latest particle accelerator) and financial assistance from private donors. Work continued for a year and a half on an NMR machine with a superconducting magnet large enough to accommodate a human body. In July 1977, Damadian and his students succeeded in creating a crude image of the cross section of a human chest, accomplished by moving the subject through 106 slightly different positions to build up an image. (The focal point was achieved through a combination of electrically focusing the RF field and taking advantage of inhomogeneities in the magnet's main field.)

Nine months after achieving the chest image, Damadian left the university to set up FONAR Corporation. He and a staff of about 30, working in a rented facility, produced a permanent-magnet MRI machine and introduced it at the 1980 meeting of the American Roentgen Ray Society and, later that year, at the annual meeting of the Radiological Society of North America. Subsequent models of the FONAR machine adopted Lauterbur's projection method, but that approach was quickly superseded by the pulsed Fourier transform method that became the dominant design of the MRI industry.

Paul Lauterbur received a B.S. in chemistry from Case Institute in 1951. He took a job with the Mellon Institute in Pittsburgh and, after serving in the Army, returned to Mellon in 1955. While in the Army, he helped set up an NMR laboratory and began research on NMR spectroscopy. He then set up his own NMR lab at Mellon and produced numerous papers. He worked on his Ph.D. in chemistry at the University of Pittsburgh while at Mellon, receiving the degree in 1962. He moved to SUNY Stony Brook's chemistry department in 1963 and continued his NMR studies, focusing on intramolecular and solvent isotope effects on chemical shifts. Lauterbur says that two areas that began to interest him at Stony Brook led toward imaging. One was biological applications of NMR, which led him to do a sabbatical in 1969-70 at the Stanford Medical Center, where he worked on labeling of proteins and tritium NMR. The second was computer-aided acquisition and processing of NMR spectra. Because the Stanford experiments were not productive, Lauterbur had difficulty obtaining support for his subsequent research (most proposals were submitted to NIH).

Through an unusual confluence of events, Lauterbur ended up at NMR Specialties in the spring of 1971, observing researchers from Johns Hopkins attempt to replicate Damadian's results. While at Mellon, Lauterbur had helped to set up NMR Specialties and, as a result, was on its Board of Directors in 1971. Because the company was about to go bankrupt, Lauterbur agreed to take over the company (a decision made easier because he did not have summer salary or grant support that year). At NMR Specialties that summer, Lauterbur watched as Leon Saryan from Johns Hopkins used the same machine Damadian had used to compare the relaxation times of fast- and slow-growing tumors in rats with those of normal rat tissue. Immediately afterward, Lauterbur went out for a hamburger, thinking about how the information obtained from invasive techniques might be obtained in other ways, so that the location of a signal within an object might be identified. That evening, he struck on the idea that inhomogeneous magnetic fields introduced locational coordinates into NMR signals, and immediately bought a notebook, wrote the ideas down, and had them witnessed.

During the next several days, Lauterbur figured out how to create a series of one-dimensional projections by changing the orientation of the gradient field and then mathematically reconstruct a two-dimensional image. (Lauterbur was not aware of previous techniques for accomplishing this, nor did he know that the principle was being applied to CT scanning at about this time.) He attempted to patent the idea privately but failed, and subsequently (when back at Stony Brook) could not arouse the interest of Research Corporation, which handled intellectual property rights at Stony Brook (SRI interview, April 12, 1996). He turned to experiments at Stony Brook in 1972, supported by part of an institutional research grant to Stony Brook from NIH. Lauterbur used a pair of capillary tubes filled with water and D₂O in one experiment and water and a solution of MnSO₄ in water in a second. The results were fuzzy but recognizable pairs of images of the cross-sections of the tubes, images that changed as a result of the effect of different RF power levels and T₁ relaxation times of the different liquids. (His initial results were obtained by hand on graph paper rather than on a display tube. A typewriter typed out the 400 spin density figures in a 20 20 matrix.) The paper he submitted to Nature describing the experiment was published in 1973 (Lauterbur, 1973). The paper explicitly mentions the possible application of NMR imaging to the "in vivo study of malignant tumors, which have been shown to give proton nuclear magnetic resonance signals with much longer water spin-lattice relaxation times than those in the corresponding normal tissues." At about this time, Lauterbur obtained a grant from the National Cancer Institute that enabled him to buy an NMR spectrometer and develop a research team (Mattson and Simon, 1996: 722).

After his initial publication, Lauterbur gave numerous talks in universities and industry, but aroused little interest in the latter. He visited the United Kingdom and interacted with the Nottingham physicists, specifically Hinshaw, Moore, Andrew, and Mansfield. He also visited Hounsfield at EMI. He traveled to the United Kingdom frequently because he received more positive reactions from the British researchers, who were convinced something important would come from their work. He consulted with industry and was visited in his lab by industry. No one treated his work seriously from a commercial point of view; he tried to sell his ideas at Varian but was unsuccessful. Technicare came to his lab at Stony Brook, but nothing came of it. As Chen and Hoult point out,

Curiously, with the exception of the EMI company in Britain, ... industry took little or no notice of Lauterbur's invention, and it was left to universities to develop magnetic resonance imaging. (Chen and Hoult, 1989: 40)

During the next 10 years, Lauterbur and his students and colleagues conducted studies of the chemical shift imaging of proton signals, the introduction of paramagnetic contrast agents, and three-dimensional projection reconstruction and, toward the end of this period, a whole-body 3D system in which cancer specimens that had been removed through surgery was studied (Mattson and Simon, 1996: 727). In 1985, he left Stony Brook to join the faculty of the University of Illinois at Urbana-Champaign, where he is now Director of the Biomedical Magnetic Resonance Laboratory.

NSF Role. Our interview with Lauterbur revealed that, for most of the period during which he was building the basis for his fundamental contribution, he struggled to obtain support, mostly from NIH. As we saw, at the time of his initial experiments, he was working with a small amount of money from Stony Brook that in turn came from an NIH institutional grant. His first interaction with NSF came in the 1980s, when he wanted to get started on nonbiological materials. He did obtain support from the RANN (Research Applied to National Needs) Program, but the program was terminated in the middle of his grant period. Our search of the NSF awards database showed that Lauterbur received a 2-year award in 1966 for $61,000, "Anisotropic Nuclear Magnetic Shieldings in Solids," and another in 1980 for $98,000 for 30 months, "Nuclear Magnetic Resonance Microscopy." A search of the database using the keywords "magnetic resonance imaging" yielded just two awards during this period: $309,000 to Jerome Singer[52], an electrical engineer at the University of California-Berkeley, in 1977 for "In-Vivo Nuclear Magnetic Resonance Imaging of Flow Patterns" and $108,000 to Tara Das at SUNY Stony Brook in 1987 for "Theory of Relaxation Times Associated with Contrast Agents in Magnetic Resonance Imaging." A search of the database using "magnetic resonance" yielded more than 900 small research awards from 1955 to the present, totaling more than $90 million. A large proportion of these awards were for the purchase of NMR spectrometers. The NIH awards database, which is accessible on-line, shows that Lauterbur was awarded his first NIH grant in 1972, and from then on averaged about two a year through 1984. In 1974, he received $232,000 from NIH for "Application of NMR Zeugmatography in Cancer Research." [53]

At SUNY, following publication of his NRM experiments with rat tumors, Damadian received grants from the American Cancer Society, the New York City Health Research Council, and private philanthropists. Long Island investors provided the capital to help start FONAR (SRI interview with Damadian, August 17, 1996). A search of the NSF database confirmed that Damadian has not received support from NSF.

Pulsed Fourier Transform Methods

Richard Ernst, a Swiss citizen, received his Ph.D. in chemical engineering from the Federal Institute of Technology (ETH) in Zurich in 1962. Upon graduation, he came to the United States to accept a position at Varian Associates in Palo Alto, where he worked with Weston Anderson, who had been a student of Felix Bloch's at Stanford. Ernst and Anderson developed the Fourier transform technique for generating a frequency spectrum output from NMR as a means of increasing the sensitivity of the instrument. The paper announcing their results was published in 1966 (Ernst and Anderson, 1966: 93). After realizing the significance of their work in 1965, Anderson and Ernst filed for a patent, "Impulse Resonance Spectrometer Including a Time Averaging Computer and Fourier Analyzer," which was granted in 1969. [54] When he returned to Switzerland and ETH in 1968, he continued his FT work. Most of the work continued to explore time-domain FT, which led to the development of 2D and 3D Fourier transform techniques. His work on 2D FT methods was central to his winning the Nobel Prize.

Ernst realized that FT methods could be applied to NMR imaging. Kumar and Welti, in Ernst's laboratory, carried out the first experiments (Kumar, Welti, and Ernst, 1975). Mansfield, at Nottingham, improved on Ernst's work by recognizing that a sequence of pulses could be used (Mansfield, Maudsley, and Baines, 1976). Then Ernst developed a 2D Fourier transform method, which was implemented by Edelstein, Hutchison, and Mallard at Aberdeen and called the spin-warp method (Edelstein, Hutchison, Johnson, and Redpath, 1980: 751). This sequence of events, based primarily in universities in Switzerland and the United Kingdom, produced the computer-driven, complex pulsed Fourier transform methods that are used in commercial medical MRI machines. Ernst has pointed out that these developments, especially the implementation of his Fourier transform method, would not have been possible without "the introduction of inexpensive computers in the late 1960s and by the development of the fast Fourier transform (FFT) algorithm" (Ernst, Bodenhausen, and Wokaun, 1987:4; quoted in Mattson and Simon, 1996: 595).

In the case of NMR and MRI, knowledge transfer from the university to industry occurred largely through people transfer. Ernst developed FT techniques while at Varian and continued to consult with Varian after he returned to Switzerland. His patents on two-dimensional Fourier transform NMR and Fourier transform NMR image reconstruction were assigned to Varian (Mattson and Simon, 1996: 597). Technicare hired Hinshaw from Nottingham. GE hired Paul Bottomley from Nottingham and Edelstein from Aberdeen.

William Edelstein received his Ph.D. in physics from Harvard, where he worked under Robert Pound, a specialist in NMR (although Edelstein's thesis did not involve NMR). He had difficulty finding a job in the United States and took a postdoctoral position at the University of Glasgow with a group trying to detect gravitational waves. Three years later, he took another postdoc at the University of Aberdeen to work on the NMR imaging project, although he had had no previous experience with NMR. In all, Edelstein spent more than 6 years in the United Kingdom, and he and Bottomley, an Australian who had worked in the United Kingdom and then as a postdoc at Johns Hopkins, both joined the GE Corporate Laboratory in Schenectady in 1980. Thus, Edelstein arrived at GE "equipped with his spin-warp NMR imaging method and experience from constructing a 0.04 T whole body imager" (Bottomley, 1996: 237). According to Edelstein, Ernst's work was the antecedent of spin-warp, but Edelstein had not actually read Ernst's papers and only later realized their relevance (SRI interview with Edelstein, April 15, 1996). Edelstein described his early work in the United Kingdom as government-funded, university-based research. He estimated that the Aberdeen group spent a total of about £200,000 on the project during his 3-year stay there.

At the time Edelstein and Bottomley joined GE, there was little business commitment to imaging. Edelstein's boss, Roland Redington, had tried to get GE Medical Systems in Milwaukee interested in imaging but was unable to get Medical Systems funding. Redington and his colleagues in Schenectady then decided to take a more research-oriented tack and obtained corporate funds to buy a 1.5 T, whole-body, high-field magnet to try in vivo NMR spectroscopy. According to Walter Robb, who was Vice President and General Manager of Medical Imaging at the time, GE was interested in MRI, but the research undertaken was not very serious because "GE was very busy with CT" (SRI interview with Walter Robb, March 19, 1996). Bottomley, defying the prevailing consensus that was partly based on one of his own earlier papers, led the GE Schenectady group in a successful effort to make head images at 1.5 T. These images "stunned" attendees at the RSNA (Radiological Society of North America) meeting in 1982, the biggest radiology meeting of the year, leading customers to defer purchases of MRI equipment until they could see what GE would offer. Under the leadership of Robb, GE Medical Systems then decided to create and market a 1.5 T product, which required a number of inventions, notably the "birdcage" RF coil by Cecil Hayes at GE Medical Systems. All of the GE development was supported by internal GE funds (Edelstein, personal communication, October 28, 1996).

NSF Role. The NSF awards database shows no direct support for any of the principals involved in originating or developing pulsed Fourier transform methods for imaging. Edelstein recalls that his Ph.D. work at Harvard was supported by the Office of Naval Research. His adviser, Pound, had received, and continued to receive, substantial support from NSF. According to the awards database, between 1970 and 1980, Pound received several awards totaling nearly $400,000 for "Resonance and Radiation Physics," a $300,000 grant for "Nuclear Resonance," and a $7,000 grant from the Instructional Scientific Equipment Program.

At General Electric, MRI development was based almost entirely on internal resources, once the British-honed talent had been imported. During development of CT scanners, GE had worked with the University of California, San Francisco, on a research level, but UCSF had joined with Diasonics, a competitor in MRI. Because of the profits generated from GE's CT scanners, there was no shortage of resources and no need to obtain outside help. Collaboration with the University of Pennsylvania, Yale, Duke, and the Medical College of Wisconsin was strictly for clinical testing of new MRI units. GE did not obtain significant technical help from universities, although physicians at clinical sites provided considerable feedback and developmental help. [55] Walter Robb had met with Damadian to discuss MRI, and GE had hired Lauterbur as a consultant. Robb recalls no NSF involvement in the development process (SRI interview with Robb, March 19, 1996).

RF Coil Design

The literature does not point to a clear breakthrough in RF coil design analogous to, say, the Fourier transform technique for image analysis. Rather, there appear to be several possible designs that have been developed over the years by both researchers and commercial firms (Chen and Hoult, 1989: 150). Gary Glover, Professor of Radiology at the Stanford Medical School, credits Cecil Hayes with a significant advance: the "birdcage" coil design (SRI interview with Glover, January 31, 1996). At higher field strengths, the body is a dielectric absorber, and as General Electric went to 1.5 T machines in the early 1980s, this fact became a significant problem. The birdcage RF coil, developed at GE, enabled whole-body imaging at these field levels.

In the early 1980s, Diasonics and Technicare were the technical leaders in MRI. Following GE's unsuccessful attempt to buy EMI Technology in 1980, GE launched an all-out effort to develop MRI (Walter Robb, personal communication, October 8, 1996). GE's Walter Robb saw the work of the company's competitors at the first annual meeting of the Society of Magnetic Resonance in Medicine and decided immediately to develop a machine. By the November 1982 meeting of the Radiological Society of North America, GE was able to demonstrate head images from a 1.5 T machine that were far superior to any competitor's. The details of the birdcage coil were kept secret for 3 years, until Hayes and his colleagues published a paper describing the design (Hayes, Edelstein, Schenk, Mueller, and Eash, 1985). By this time, GE had established its place as the market leader (SRI interview with Hayes, April 3, 1996). Gary Glover, a researcher at GE Medical Systems during this period, noted that "GE was in considerable collaboration with university researchers at Duke and Stanford. There was much infusion from the university folks. The university researchers would propose a design, and a company would build it to try it out." Still, much of the innovation required to make MRI commercially viable for clinical applications resulted from research in industrial laboratories (SRI interview with Glover, January 31, 1996, and personal communication, October 1996).

Cecil Hayes studied physics at Harvard during the mid-1960s; he and Edelstein both had Robert Pound as their thesis adviser. He took a 3-year postdoc position at Rutgers, then a second postdoc at the University of Utah. He became involved in NMR imaging about 1978 while at Utah, when several NMR physicists joined a group of electrical engineers and pulmonary M.D.s, who had previously tried to do microwave measurements of water in the lungs. Hayes focused on hardware, building a small NMR unit from scratch. In 1982, he joined General Electric, where his fellow student Edelstein was working at corporate R&D headquarters in Schenectady. Hayes applied to the R&D lab but was hired by the manufacturing facility in Milwaukee (SRI interview with Hayes, April 3, 1996). Hayes is now a member of the department of radiology at the University of Washington.

NSF Role. The first 2 years of Hayes' graduate study were paid for by an NSF Predoctoral Fellowship. Hayes recalled that his thesis advisor, Pound, had an NSF grant, which supported Hayes for several years. His postdoc at Rutgers and the first 3 years of his second postdoc at Utah were also based on NSF support for NMR applications to condensed matter. The MR imaging work at Utah was supported by NIH. His work at GE was supported entirely by internal company funds (Hayes, personal communication, October

CONCLUSIONS

Government, Industry, University Roles and Relationships

For the first decade of its development, MRI research was led by academics, many based in the United Kingdom, who were either physicists interested in the imaging phenomenon itself or medical researchers interested in eventual application of the technology to clinical diagnostics. Initially publishing in physics journals, by the late 1970s most of the British were publishing in radiology and instrumentation journals, driven by the prospect of the technology's application in clinical medicine. In the mid-1970s, only EMI pursued industrial development of MRI, but it did so substantially in isolation from other researchers. The British government was substantially involved in MRI development; British researchers received significant support from the Medical Research Council.

In the United States, Raymond Damadian, a medical researcher, sparked interest in applying NMR techniques to medicine, while Paul Lauterbur, a chemist, initiated imaging itself and pursued improved images. [56] Until well into the 1980s, both worked essentially alone, at least as far as colleagues in the United States were concerned. Damadian pursued commercialization of MRI on his own, while Lauterbur unsuccessfully tried to interest U.S. industry in its development. Once the potential for applying MRI to medicine became apparent to industry, largely because of the clinical trials being conducted in the United Kingdom with machines built by academics, industry rapidly took over the leading role and kept it through the end of the 1980s. Rapid commercial development and market leadership were achieved by large medical instrument firms, such as General Electric and Siemens, who used profits, technology, and market knowledge from related products to move quickly to the forefront. In most cases, companies took the lead in research underlying MRI rather than relying on collaborations with or results generated by universities. Some academic consultants were hired, but the primary industry-university linkages (which were important) consisted of companies working with medical schools to conduct clinical trials of prototype machines. GE and Technicare, an early technical leader in the field, basically bought their knowledge and expertise from the British in the form of the leading researchers: Edelstein, Bottomley, Hinshaw. It is significant, of course, that Edelstein and Hinshaw were Americans, educated in the United States.

Relationships between Fundamental Research and Technology Development

The fundamental elements of MRI are those of NMR, which rest directly on nuclear physics. NMR moved rapidly from fundamental research at Stanford to commercialization because of Russell Varian's involvement; from that point, industry led the development. Richard Ernst's initial contribution to NMR came while he was at Varian. Fast Fourier transforms, developed by mathematicians at Bell Labs, and Ernst's 2D imaging technique based on pulsed Fourier transform procedures, speeded the development of MRI by reducing computing requirements. Once the phenomenon of nuclear resonance was understood, the development of both NMR and MRI depended more on technology than on research, although as commercial NMR matured in the mid-1980s, the focus of advances shifted more toward fundamental research. [57] Through most of MRI's evolution, the challenges were to improve image quality and reduce scan times and computational requirements. Even during the early days of MRI at Nottingham, Aberdeen, and Stony Brook, these challenges, rather than a desire to explore underlying processes, motivated the researchers-yet in many instances it was necessary to engage in just such explorations to improve image quality or reduce scan time.

Intellectual Property Rights

There is an extensive patent history associated with MRI, but there is no evidence that patenting slowed development of the innovation. The pioneers of MRI, academics in the United States and the United Kingdom, either recognized the potential value of their discoveries or hedged their bets by patenting. Lauterbur attempted unsuccessfully to patent his projection method. Damadian patented the FONAR technique. Hinshaw and Mansfield patented the point and slice methods of selective excitement, respectively. Ernst, after returning to the Swiss Federal Institute of Technology, patented his 2D technique and assigned the patent to Varian. The leading companies patented extensively; the following table shows the number of MRI patents granted to selected firms in the MRI industry between January 1, 1971, and December 31, 1990. [58]

An examination of the patent citations shows that there was clear awareness of others' previous work. Although researchers may have proceeded with prototype development before addressing the intellectual property question, most subsequently took out licenses or engaged in cross-licensing with other firms. GE and other companies all licensed the portfolio held by the British Technology Group. Hinshaw commented on the tradition in the medical instruments industry of using each other's patented technology and competing on other grounds to avoid patent litigation. It appears that, with some notable exceptions, such as GE's birdcage coil, market advantage was gained more through know-how and pricing than through intellectual property, since most companies were using the same basic technology.

There is more to the property rights story than we were able to capture in this case. Unfortunately, the detailed bibliometric and patent citation analyses that were to have accompanied this case were not funded until the case was completed; therefore, whatever insights these analyses will yield will have to be added later. In any event, a more detailed examination of the key patents in MRI and an exploration of patenting strategy by firms seem warranted.

NSF Role

Through awards for research and instrumentation in nuclear magnetic resonance, NSF provided a significant part of the infrastructure that several U.S. participants drew on in the development of MRI. From 1955 to the present, NSF support for NMR research underlying the entire innovation amounted to $90 million. [59] Among those receiving awards were Harvard's Robert Pound, dissertation adviser to both William Edelstein and Cecil Hayes. In addition, NSF supported Hayes' coursework, his dissertation work, and 6 years of postdoctoral work at Rutgers and Utah. This included support for instrumentation purchases and the education of graduate students, some of whom would build on their graduate research experiences in subsequent MRI work. Additionally, NSF has supported research in related areas undergirding the innovation itself, e.g., electromagnetics, digital systems, and computer engineering. Beyond this, however, there is little additional evidence of NSF's direct role in the development of MRI, at least through the end of the 1980s. In addition to small research awards, NSF provides partial support for research on MRI imaging at the NSF Engineering Research Center for Emerging Cardiovascular Technology at Duke and, until recently, did so for the Center for Magnetic Resonance Technology for Basic Biological Research at the University of Illinois. [60]

REFERENCES

Andrew, E. Raymond. "NMR Imaging," Accounts of Chemical Research, 16 (1983): 114-122.

Blume, Stuart S. Insight and Industry: On the Dynamics of Technological Change in Medicine. Cambridge, MA: MIT Press, 1992.

Bottomley, Paul A. "The Development of High-Field NMR Imaging: 0.12 T to 1.5 T," in Grant and Harris (1996).

Chen, C-N, and Hoult, D. I. Biomedical Magnetic Resonance Technology. New York: Adam Hilger, 1989.

Cooley, J. W., and Tukey, J. W. Math. Comput., 19 (1965).

Damadian, R. V. "Tumor Detection by Nuclear Magnetic Resonance," Science, 171 (March 19, 1971): 1151-1153.

Damadian, R., Goldsmith, M., and Minkoff, L. "NMR in Cancer XVI: FONAR Image of the Live Human Body," Physiological Chemistry and Physics, 9 (1977): 97.

Damadian, R., Minkoff, L., Goldsmith, M., Stanford, M., and Koutcher, J. "Field Focusing Nuclear Magnetic Resonance (FONAR): Visualization of a Tumor in a Live Animal," Science 194, (1976):1430.

Edelstein, W. A., Hutchison, J. M. S., Johnson, G., and Redpath, T. Physics in Medicine and Biology, 25 (1980).

Ernst, R. R. "Nuclear Magnetic Resonance Fourier Transform Spectroscopy," Angewandte Chemie, International Edition, 31, 7 (July 1992): 805-930.

Ernst, R. R., and Anderson, W. A. "Application of Fourier Transform Spectroscopy to Magnetic Resonance," Review of Scientific Instruments 37 (1966).

Ernst, R. R., Bodenhausen, G., and Wokaun, A. Principles of Nuclear Magnetic Resonance in One and Two Dimensions. Oxford: Clarendon Press, 1987.

Feeney, James. "Development of High Resolution NMR Spectroscopy as a Structural Tool," in R. Bud and S. Cozzens, eds., Invisible Connections: Instruments, Institutions, and Science. Bellingham, WA: SPIE Optical Engineering Press, 1992.

Grant, D. M., and Harris, R. K., eds. Encyclopedia of Nuclear Magnetic Resonance, Volume I: Historical Perspectives. New York: Wiley, 1996.

Hayes, C. E., Edelstein, W. A., Schenk, J. F., Mueller, O. M., and Eash, M. Journal of Magnetic Resonance, 63 (1985): 622.

Hinshaw, W. "Notes on the History of MR Imaging from My Perspective," in Grant and Harris (1996).

Hinshaw, W. "Spin Mapping: The Application of Moving Gradients to NMR," Physics Letters, A48 (1974): 87.

Hinshaw, W. S., Bottomley, P. A., and Holland, G. N. ""Radiographic Thin Section Image of the Human Wrist by Nuclear Magnetic Resonance," Nature, 270 (1977): 722.

Hinshaw, W. S., et al. "Display of Cross Sectional Anatomy by Nuclear Magnetic Resonance Imaging," British Journal of Radiology, 51 (1978): 273.

Hollis, Donald P. Abusing Cancer Science, Chehalis, WA: Strawberry Fields Press, 1987.

Jackson, J. A., and Langham, W. H. Review of Scientific Instruments, 39 (1968).

Kumar, A., Welti, D., and Ernst, R. R. Journal of Magnetic Resonance, 18 (1975): 69.

Lauterbur, P. C. "Image Formation by Induced Local Interactions: Examples of Employing Nuclear Magnetic Resonance." Nature, 242 (1973): 190-191.

Lauterbur, Paul C. "Cancer Detection by Nuclear Magnetic Resonance Zeugmatographic Imaging," General Motors Charles F. Kettering Prize Lecture, Cancer, 57 (1986): 1899-1903.

Lunzer, Francesca. "Shakeout in Medical Scanners," High Technology Business (June 1988): 37.

Mansfield, P., and Grannell, P. K. "NMR 'Diffraction' in Solids?" Journal of Physical Chemistry, 6 (1973): L422-L426.

Mansfield, P., and Maudsley, A. A. "Medical Imaging by NMR," British Journal of Radiology, 50 (1977): 188.

Mansfield, P., Maudsley, A. A., and Baines, T. "Fast Scan Proton Density Imaging by NMR," Journal of Physics (E): Scientific Instruments, 9 (1976): 271.

Mansfield, P., et al. "Human Whole Body Line Scan Imaging by NMR," British Journal of Radiology, 50 (1978): 188.

Mattson, James, and Simon, Merrill. Pioneers of NMR and Magnetic Resonance in Medicine. Jericho, NY: Dean Books, 1996.

Morris, Peter. "Present and Future Applications of NMR to Medicine and Materials Science," R. Bud and S. Cozzens, eds., Invisible Connections: Instruments, Institutions, and Science (pp. 223-225). Bellingham, WA: SPIE Optical Engineering Press, 1992.

Odeblad, E. Z. Med. Isotopenforsch., 1, 25 (1956).

U.S. Congress, Office of Technology Assessment. Policy Implications of the Computed Tomography (CT) Scanner: An Update. Background Paper. Washington, DC: Office of Technology Assessment, January 1981.

U.S. Congress, Office of Technology Assessment. Nuclear Magnetic Resonance Imaging Technology: A Clinical, Industrial, and Policy Analysis. Washington, DC: Office of Technology Assessment, September 1984.

SRI INTERVIEWS

Gary Glover, Stanford University

Walter Robb, General Electric (retired)

Paul Lauterbur, University of Illinois

C-N Chen, National Institutes of Health

William Edelstein, General Electric

Raymond Damadian, FONAR Corporation

Cecil Hayes, University of Washington