Paul Lauterbur and the Invention of MRI

by M. Joan Dawson

  Changes in relaxation times are not, by themselves, diagnostic of cancer. But Damadian started a productive ferment, and there was a lot of excitement about this possibility during the early 1970s. Dr. Damadian believes he has not been given nearly enough credit for his work during this period. He campaigned vigorously for the Nobel Prize for many years. When the Nobel Prize in Physiology or Medicine was announced in 2003, for the invention of MRI, he placed full-page advertisements about injustice in major newspapers around the world showing the Nobel medal upside down.

  Yes, his work was important, to the point that Paul conceived of MRI as a result of watching a repetition of Damadian’s experiments. But his claim that pathology and relaxation behavior could have a numerical correlation was soon debunked. Damadian was nowhere near the first person to suggest the medical uses of NMR, as he claimed. That honor probably belongs to Erik Odeblad of Sweden, who in 1949 started studying human tissues while he was with Purcell’s group at Harvard and later continued in Stockholm. For some reason, while this group studied many tissues, they did not look at cancers.

  Damadian often claimed (but, curiously, sometimes denied) that he invented MRI. According to Damadian’s biographer, Sonny Kleinfield,

  Unlike Lauterbur, however, Damadian was not at this time [1972] thinking about making pictures with an NMR machine. Imaging, specifically, had not occurred to him, and he would not begin to contemplate its importance until Lauterbur proposed his technique. Damadian was mostly interested in gathering chemical data that would reveal diseased states; he viewed imaging, when he began to think about it, as just one of the subsidiary usages of the chemical maps obtained by NMR scanners.1

  The distinction here between an idea that it would be nice to do body scans to diagnose cancer and the physical reality of imaging is subtle enough to confuse people who were not intimately involved in these events. But it makes all the difference in the world, like the difference between air and rock crystal. It never occurred to Damadian to use magnetic field gradients, and without field gradients you don’t have MRI. It is worth repeating: MRI is completely, entirely, dependent on the use of field gradients. It may be that Damadian never appreciated the importance of field gradients once Paul had pointed this out; this would explain a lot of Raymond’s anger at not being credited with imaging.

  Paul invented a generally practicable technology. Damadian had an idea related only to cancer relaxation times—important enough, but not imaging. Also confusing is that Damadian produced a sketch of a possible NMR system for detecting cancer in vivo in a patent filed in 1972.2 This, he sometimes claims, gives him priority in the invention of MRI. The trouble is that Damadian’s cancer scanning machine of 1972 was based on an incorrect understanding of physics and could never be implemented. It depended on crossbeams of a special magnetic field and a radio-frequency field. Well, you cannot make radio-frequency fields into a beam like a flashlight, and magnetic fields cannot be focused in this way. The method was never used either for scanning cancerous tissue or for imaging because it was impossible to do so.

  And how, then, was he to obtain those chemical maps? In 1977 he used a technique he named FONAR, which called for moving a patient back and forth across a “focused” magnetic field created by destroying the field in all but one spot (in his imaging spectrometer that he called Indomitable, which was picked up by Sonny Kleinfield as the title of his biography of Damadian). This was explained in a public relations effort by SUNY Downstate Medical Center: “The proposed NMR device for detecting cancer in humans would not have to be highly elaborate, Dr. Damadian says. It would consist of a large coil to emit radio waves and a movable magnet to create the magnetic field required. The coil would be wrapped around the patient’s chest, while the magnet passed back and forth across the body. A detector would pick up NMR emissions for analysis.”3 There are two problems here. Focused detection, which Damadian was describing, was not novel, and the method was a blind alley both because it simply takes far too long to be practicable and because there is no way to develop it further. All MRI is now done using variations of Paul Lauterbur’s and Peter Mansfield’s techniques, that is, field- and time-dependent gradients.

  People love scientific feuds, and the idea of fisticuffs between Lauterbur and Damadian has received wide press. Damadian’s feelings are laid bare in Kleinfield’s biography of him: “After reading Paul Lauterbur’s Nature paper, Damadian ‘blew up.’ ‘Here I was talking about medical scanning and getting ridiculed and here was this guy standing up and saying that he had invented it. I was absolutely shocked. I couldn’t believe it.’”4 Kleinfield said of Damadian, “The memory of the incident firmly screwed itself into a corner of his mind. It would torment him for years, heighten his nervousness, convinced him that he was a target of organized persecution. A bad turn of events and some voice deep inside him would whisper that somehow Paul Lauterbur was behind it.”5

  Paul reacted by not commenting—for almost forty years. It was especially difficult when Damadian made outrageous claims, and reporters were suspicious when Paul would not discuss the matter. The one comment I have found is in a private letter to another scientist, for whom Paul annotated several pages of Kleinfield’s book. This passage appears on page 55: “Over the years the two have fired a good many astringent salvos at each other.” Paul’s marginal annotation was, “Not by me!” Other scientists in the field generally refused to talk publicly about the Damadian controversy, but some leaders did have explicit things to say.6 Don Hollis went so far as to write a book about it, Abusing Cancer Science.

  The Zeugies

  They were young, inexperienced, and inspired. It’s hard to overstate the excitement of Paul’s group at Stony Brook in the early 1970s and the intense camaraderie that connected these young people. They threw themselves fully into their work and knew they were making history. They worked hard together, and they celebrated each other’s birthdays and graduations. The first students to demonstrate the potential of MRI were undergraduates—Chuck Dulcey, Cliff Weizel, and Fran Predo—and Mike Feiler, who had a BA in English and was then taking prerequisite courses for medical school. They worked for academic credit and tiny stipends. In 1973 Paul was able to hire graduate students and postdoctoral research associates as well.

  C.-M. Lai, a postdoctoral fellow with a physics background, was the first to arrive. A very serious, studious, and disciplined person, C.-M. took over the lab in an experimental sense, helping the junior students with their projects. He had been a graduate student at Brookhaven National Laboratory but got bored. He switched institutions and majors to join Paul’s group.

  Graduate students David Kramer and Reginald Dias and postdoctoral fellow Waylon House also joined Paul when he finally received funding to support them. The small laboratory was quickly joined by F. W. Porretto, a computer specialist, postdoctoral fellow Ching-Nien Chen, and, for a short time, by M. J. Jacobson, a thoracic surgeon on leave from his position with the Veterans Administration. Another early member of the laboratory was Joe Frank, who started working in the lab as a photographer and, becoming excited about the project, joined as an undergraduate research assistant and continued as a graduate student. They called themselves the “Zeugmatographers,” or “Zeugies.” They wore lab coats with a “Z” emblazoned on the arm bands and t-shirts with a large black Z on the front, in imitation of Superman.

  Figure 7.1

  Paul receives a site visit. Pictured are members of Paul’s lab and site visitors sent by the National Institutes of Health to evaluate the lab’s performance.

  These first students explored as many different imaging methods as possible, and it is astounding how many different directions they pioneered. They worked together and learned from each other, under Paul’s overall guidance. The first objects to be imaged were chosen in part for their simple structures, which were not too taxing for the laboratory’s evolving methodology, and also for their size. Paul’s first work on capillary tubes and miniature clams had been done using an o
rdinary commercial NMR spectrometer that could accommodate objects no larger than 4 mm in diameter. He was a long way from the goal of human imaging.

  By 1973 he had modified an instrument, a Varian DP-60, with a larger magnet gap. He came by this because Fairchild Industrial Products closed its laboratory on Long Island. Paul happened to know their NMR man, and the machine was donated to him. He and his students fitted together five crates worth of magnets, which had been shipped in pieces. They wound coils used to direct the magnetic energy, built their own radio receiving and transmitting coils, wrote computer programs, and hooked up a TV monitor to their contraption. Now pecans (the oil of the nutmeat was imaged rather than the water) were accessible, and so were a pine branch and the larger, delectable, cherrystone clams. They had their first patient noncompliance problem when a mouse refused to enter the sample tube. Mouse tails were big for a while and, a great milestone, the thorax of a mouse. Everyone remembers that mouse thorax well. It was more complex and biologically interesting than any of their other images, and they came away thinking, “Now we have seen the future.”

  The group imaged eggplants, sweet peppers, and oranges, and then bigger objects such as coconuts and pork chops, all simple biological specimens that could be cut open to compare the image with the real thing. A detailed image of a green pepper caught the attention of everyone in NMR because its internal structure was clearly delineated. Paul’s postdoctoral fellow, Waylon House, explained: “We were hot. It was exciting. I had never seen anything before the green pepper that was really good, that was anything more than a blob.”7 Each new specimen was used to investigate further the characteristics of, or to solve problems inherent in, the imaging techniques to be used for human subjects and patients. They were then often eaten for lunch.

  Big Red

  Paul had calculated that human-scale magnets were possible, much to the skepticism of his colleagues. By 1975 his laboratory couldn’t make much more progress without a bigger magnet, preferably a human-sized magnet. To get one, he needed money. Paul finally raised funds to develop MRI technology from the National Cancer Institute, a part of the National Institutes of Health, and he ordered a magnet such as had never been built before. He was able to talk Walker Scientific, a company that had been in magnetics since the late nineteenth century, into providing a human-sized magnet for the money he had available, but severe compromises had to be made.

  Medical MRI systems now typically have a 1-meter (33 inches) or larger bore. Or they have open or stand-up spaces with no tube at all. Paul initially asked for a magnet having a 60 cm (28-inch) bore, large enough to accommodate a sizable portion of the American population and good enough for the development work he had in mind. The magnet he received had a 45 cm bore (18 inches), and when he complained about the quality of the coils used to transmit and receive NMR signals, these were improved but made bigger, so he ended up with 42 cm (16 inches) of usable space, which will accommodate practically no one. This was the first one of the kind that Walker Scientific ever built, so perhaps it is not too surprising that there were bugs.

  Figure 7.2

  Big Red under construction.

  Paul had a cartoon tacked to a wall in the lab. It showed a newly built canoe, with one end facing upward and one end facing downward. A man sits at a desk with his head in his arms, and he appears to be on the verge of tears. Underneath is scribbled one word: SHIT! Paul told Sonny Kleinfield, “This pretty much expressed our feelings when the machine came in.”8 The whole laboratory was disappointed that they could not do human imaging but knew there was a great deal of other important work to do to make MRI clinically useful. They began using this system for scaling things up. Damadian, who believed that to do the first human image was the only possible goal, had his own ideas about Paul’s bad fortune: “I figured the good Lord had had his way of seeing justice. We had a big laugh about it for several days.”9

  It was a magnet made for a homunculus. The students painted their much-too-little magnet a bright color and called it “Big Red.” Years later it was rechristened “The little red model T.”10 There were delays getting it running (there were always unforeseen problems in getting magnets running), and poor Red had a manufacturing flaw so that insulation material flaked off into the magnet bore. The Walker salespeople were surprised: “Gee, that never happened before!” Then Paul got to know the magnet builder, who told him, “It happens to everyone.” Red lived in the basement of the chemistry building, which securely supported its weight. Metal beams ran through the floor of the basement, so Red roosted on a comfortably raised bed, free of magnetic interference.

  Paul and his little group now concentrated on working out the methodologies and showing the feasibility of various clinical applications of zeugmatography. But he had a problem: he knew nothing about medical science or clinical practice. He could therefore be seen of a morning in the new medical library at Stony Brook skimming innumerable books and journals, looking for themes and threads and picking up the vocabulary he needed to talk to physicians. As a result of these forays, he came up with specific projects for his own laboratory and suggestions that would drive the development of diagnostic MRI for decades to come, so that other scientists would be inspired to carry on.

  Experimental Verification

  In some years there seem to have been more public presentations of research projects than there were workers in Paul’s laboratory. For example, the year 1976 brought forth four very different experimental projects with direct relevance to medical diagnostics. “In vivo Zeugmatographic Imaging of Tumors”11 deserves particular note because it followed up Raymond Damadian’s observations that changes in signal decay times accompany cancer.12 This experimental paper from Lauterbur’s group showed that if such changes existed, they would indeed be imageable by MRI. This and another paper, “Water Proton Spin-Lattice Relaxation Times in Normal and Edematous Dog Lungs,”13 were presented at the Fourth International Conference on Medical Physics, which took place in Toronto in the summer of that year. The second paper showed an even more conspicuous example of a correlation between signal decay and tissue damage.

  The same year produced “Measurement of Proton Nuclear Magnetic Longitudinal Relaxation Times” and “Water Content in Infarcted Canine Myocardium and Induced Pulmonary Injury,” both presented at a meeting of the American Federation of Clinical Research in Philadelphia.14 The dog heart study was a striking example of relaxation times correlating with tissue damage—in this case, prevention of cardiac blood flow. The group had already imaged flow in 1973,15 and predicted future MRI studies of blood flow in cardiac chambers and arteries.

  A group half a dozen strong, including the undergraduates, worked on all of these projects. In today’s large MRI laboratories at least this many full-time postdoctoral investigators might be engaged in any single one of these ongoing endeavors. Remarkably, undergraduate students did a large part of this cutting-edge research. Professors who take undergraduates into their laboratories generally don’t expect innovative ideas and activities from people still learning the ropes, and undergraduates, with their full load of classes, cannot spend a great deal of time on research. But Paul could regularly coax high-level scientific work from these undergraduates. This was true from his very earliest academic days up to the year he died—over fifty years of mentoring truly excellent young people.

  Undergraduate Chuck Dulcey, who had taken Paul’s freshman honors chemistry class, played a major role. Paul found Chuck to be a good and dedicated student and asked him if he would like to try doing some research. Chuck joined the lab in December 1972, during his sophomore year, as Paul’s first MRI student and collaborator, and stayed for two and a half years. “People can’t appreciate what it was like,” Chuck commented. “We had to figure out everything.”

  Paul had done the first tests of projection reconstruction by digitizing with his fingers. Chuck’s job was to computerize the process. His computer had 14K of memory, which wasn’t large enough to hold both the pr
ograms and the computations, so he had to load and reload the programs, and he could only use integers. Paul told him to start not with experimental data but with an array of arbitrary numbers so he could compare the output with the numbers he had put in. Using thirty-two bins, Chuck projected back the image and then recalculated, over and over again. He needed to calculate how much signal from each projection was in each pixel, and then assign the partial intensity to that pixel. People are now familiar with pixilated images with their bumpy little squares; they are fun and weird, but you need a little mental gymnastics to figure out what you are seeing.16 Chuck applied a smoothing function, Gaussian multiplication, to show how a real image might look.

  His next step was to adapt a published image-processing algorithm to run on the Stony Brook mainframe computer.17 Luckily, Chuck had taken Fortran in high school, and he was familiar with the computer center from a summer course he had taken there before freshman year. He developed three different versions of his image-processing program, Zeug 1, 2, and 3, using algebraic reconstruction techniques. All were in Fortran, and each was more sophisticated than the one before. Chuck entered data in the laboratory and this fed automatically to the innovative multi-user computer lab on the chemistry building’s third floor.18 The punch cards thereby generated were taken to an IBM 360, the campus mainframe computer. In later stages the campus acquired an IBM 370. These were exciting new machines at the time, though we now smile and shake our heads as we remember their limitations. Young people today find it hard to believe that serious work could be done on them. There was a lot to think about. “What did it mean to be inside a projection? Where was the noise ending up? There was controversy—was this the way to go?”19