Paul Lauterbur and the Invention of MRI

by M. Joan Dawson

  Before Peter Mansfield could make his unique contributions to his new field, another group from Nottingham jumped the gun. The first technique for producing MR images after Paul had shown its feasibility was developed by Waldo Hinshaw and Bill Moore of Nottingham. They had traveled to India for the International Conference on NMR, held in January 1974. At this conference Paul gave a talk on imaging that created some excitement and provoked quite a bit of discussion among the “Nottinghamsters” during their return flight to England. They realized that the magnetic gradients could be produced not only by applying fixed gradients in known directions but also by applying time-dependent gradients, such as oscillating gradients. Time-dependent gradients could be made to spoil the signals from all but a particular spot, where all the time dependence cancels out. They basically invented a technique they called the “sensitive point” method before they arrived home from India.2 It had the advantages of very simple data-processing requirements and uniform sampling of k-space, but the disadvantage that it was inevitably rather slow.

  Another early enthusiast was Richard Ernst of the Swiss Federal Institute of Technology in Zurich (ETH-Z). Richard (Nobel Laureate, 1992) was at the 1974 meeting of the ENC in Blacksburg, Virginia, when Paul gave one of his early presentations on MRI, including the first image of a living animal, a mouse. During the discussion Richard described the bright central artifact resulting from the back-projection method as the soul of the mouse. Richard was very excited about Paul’s new method but “couldn’t understand why Paul had not used pulse and Fourier transform NMR,” a Fourier transform MR spectroscopic method he had devised with Wes Anderson. Paul had anticipated (see appendixes A, B, and C) pulse and Fourier transform MRI but did not have the equipment needed to do such an experiment, and Ernst thought of a way of doing it that Paul never thought of. Ernst and his colleagues accomplished their two-dimensional imaging using these methods, but those attending his talk at a meeting in Kandersteg, Switzerland, in 1974 considered the work premature. Nevertheless, this was the next major advance after Paul’s ideas, and Paul himself was greatly impressed. “Sometimes even prematurely born children survive and excel,” Richard remarked.

  Learning of Paul Lauterbur’s work, John Mallard with his two postdocs, Jim Hutchison and Meg Foster, at Abderdeen decided they were well placed to pursue MRI (Jim and Meg also pursued matrimony). In short, they did the first clinical imaging (inadvertently demonstrating the method’s ability to show the pathology of edema at the neck of a mouse that had been killed to keep it still). They went on to develop a large clinical program.3 They were also responsible for the “spin warp” method, a practical application of Richard Ernst’s ideas. By 1974, then, six groups were involved in MRI, and there was an intense rivalry among the various groups to achieve notable firsts. These groups were at the laboratories of Paul Lauterbur, Raymond Damadian, Raymond Andrew, Peter Mansfield, John Mallard, and Richard Ernst. Andrew’s group split into two, and the three groups in the Nottingham Physics Department, especially Mansfield’s and Andrew’s, were soon not on speaking terms. Their squabbles were a delight to gossip about.

  Figure 8.1

  The first clinical MR magnet, Aberdeen. From J.M.S. Hutchinson, W. A. Edelstein, and G. Johnson, “A Whole-Body NMR imaging Machine,” Journal of Physics E: Science Instrumentation 13, no. 9 (September 1980): 947–955. Reproduced by permission.

  Let’s go back to Peter Mansfield, since he continued to devise new MRI methods. Peter’s approach to MRI was quite different from Paul’s. For the sake of speed, he decided to go for two dimensions. The first problem to be solved for accurate two-dimensional imaging is how to choose a two-dimensional slab from which the image would come, and ignore the rest of three-dimensional space. He thought of a more elegant way than Paul had envisioned. Mansfield and his group showed that slice selection could be accomplished by turning on a field gradient and then quickly turning it off, thus creating a pulse of linear field gradient.4 After the pulse is turned off, a gradient is applied in the y-direction. This gradient serves as phase memory. Then a third gradient is turned on and left on while the signal is sampled. This gradient serves for frequency encoding. A Fourier transform of the resulting signal gives each pixel a unique phase and frequency address. This sequence of pulses and gradient switching defines a single line in the slab. To sample the whole slab the sequence is repeated multiple times with increasing size of the y-gradient. And that became a standard method of obtaining an MR image.

  Peter Mansfield added an important additional step, which is the “spin echo,”5 and created echo planar imaging.6 In EPI the whole of k-space is acquired in one scan. This is possible because once one set of frequency information is acquired, the readout gradients can be reversed and spins will precess in the opposite direction and subsequently rephase, causing a regrowth of the NMR signal, known as a gradient echo. If the readout gradient is switched rapidly, the whole of k-space can be sampled before the signal is obliterated by T1 relaxation. In their first MR images obtained using EPI, Mansfield and co-workers were able to get in five echoes before the signal decayed to nothing. EPI is technically demanding but very rapid, and is capable of following the cardiac cycle and of dynamically imaging brain activation. EPI at first languished. Although all the scientists were very excited, the imaging companies didn’t want to do the required upgrade to machine specifications. Finally, GE and Siemens launched their EPI machines in 1993.

  It is interesting that Paul Lauterbur had thought about EPI as early as October 1971 (see appendix B), although he never followed it up.

  The Early Images

  Paul made the first image of a tumor in a live animal when, in 1976, he made a contour map of a tumor-bearing mouse.7 The first human NMR image, reported in 1976, was of a human finger and was obtained by Peter Mansfield and Andrew Maudsley at the University of Nottingham. Andrew could fit his index finger into the sample space of a conventional iron NMR electromagnet. Using a magnet with a larger sample space, 13 cm, Waldo Hinshaw and his colleagues made their beautiful image of a lemon and a clear slice through Paul Bottomley’s wrist,8 which demonstrated conclusively that MRI could be used in medical studies. The first human thorax MR image was obtained in Damadian’s laboratory using a home-built machine in 1977, followed the next year by the first MR image of the human head, obtained by Hugh Clow and Ian R. Young.9

  Mansfield and his colleagues acquired the first MR image of the human abdomen in 1977, using a new technique of their devising. They had been warned by Tom Budinger, the guru of MRI safety, that the rapid switching of gradients in this study might be perilous. But Peter Mansfield trusted his own calculations and served as his own test object. Mansfield wrote about the tension in the atmosphere.10 “I climbed into the machine and signaled to Peter [Morris] and Ian [Pykett] to push the button for a single pulse. There was an audible crack but I felt nothing. I then signaled to start the scan. . . . I was clamped in the magnet vertically in pitch darkness for 50 minutes until the procedure was completed. Our wives and fiancées were present to haul me out in an emergency.”

  Figure 8.2

  Peter Mansfield and Paul Lauterbur.

  The Raging Flood

  By the early 1980s there was such an explosive growth in MRI that it was difficult to keep track of imaging methods. There is now a complex library of over one hundred different ways to do MRI. Most methods involve a series of subtle combinations of radio-frequency pulses and switched gradients. They are devised for different purposes and have different positive and negative qualities. Some methods sample the relaxation time T1 in particular, and are especially sensitive to the differences between gray and white matter. Others are more sensitive to T2 (or T2*).11 Paul kept up with these methods as they appeared but did not contribute much to this cornucopia.

  Computed tomography (CT) was developed just before MRI, and at first was the gold standard to which MRI was compared. It was, in its time, a huge breakthrough, winning the Nobel Prize in Physiology or Medicine for H
ounsfield and Cormak in 1979. In some ways CT may have given an impetus to the development of MRI. It had already shown that two-dimensional sections through the head or body allow diagnosis in a noninvasive way. Our meetings were full of comparisons of various methods of MRI with CT, enough comparisons to make one’s head ache. Of course, MRI usually won these contests. To a large degree the pictures sold themselves; they had far better contrast than CT scans with similar spatial resolution. As described by John Mallard, there was a frenetic period of national and international invited presentations all over the world, and these early MRI investigators became used to living out of a briefcase. Editors welcomed, even solicited, their papers.12

  The multinational medical imaging companies crashed in, pouring megabucks into developing prototype machines as quickly as possible. Raymond Damadian’s FONAR Corporation sold four magnets, and then switched to a variation of the Lauterbur/Mansfield technology. At the 1981 annual meeting of the Radiological Society of North America, Diasonics displayed stunning head and body images from its new 0.35 T superconducting magnet, introducing the multi-echo, multi-slice approach that made MRI financially viable in clinical terms. General Electric’s first imager, presented to the world in 1984, had a 1.5 T magnet—four times more powerful than any others commercially available at the time. And it did both imaging and spectroscopy.

  The MRI companies did some clever marketing, and by 1983 MRI had begun to take on the dimensions of a mammoth industry. Although the U.S. Food and Drug Administration had not yet approved any of the machines, a step necessary for the systems to be commercially marketed and for hospitals to be reimbursed for scanning patients, manufactures were moving quickly to advance their products and grab marketing advantages. Even without FDA approval, companies were able to sell their machines as investigational devices whose application would not be covered by insurance. Manufacturers even gave away their scanners—or else agreed to be paid years down the road—to get them accepted into prestigious hospitals that would burnish the product. The year 1983 marked one decade after the first paper on MRI was published, and already there were thirteen companies committed to manufacturing NMR scanners. Diasonics was granted FDA market approval for its MRI systems in 1984. General Electric was thus exalted in 1985. It was a new world.

  One important name left out of the mix was that of Varian Corporation, which throughout the 1950s and 1960s had been almost the only name in NMR. Paul was particularly cross with them because when he had taken 13C NMR to them in the 1950s they would have nothing to do with it, and when he took MRI to them in the 1970s they would have nothing to do with it. So MRI, without Varian, began reaching the sick people who needed it.

  Turf Wars

  Paul loved the name “zeugmatography,” but not many other people did. It was just too unwieldy and foreign for medical and commercial circles. A simpler and more direct name was required. “Nuclear magnetic resonance imaging,” or NMRI, an exact description of what is accomplished, was proposed, but “magnetic resonance imaging,” or MRI, was finally settled on.

  There is a story that the word nuclear was dropped because of the connotations of radioactivity, which would scare the public and inhibit acceptance of the new methodology—patients would think they were about to undergo a micro-Hiroshima! However, every major hospital has a department of nuclear medicine, where micro-Hiroshimas do not intrude. In fact, the death of the “N” resulted from a great turf battle between radiology and nuclear medicine over which discipline and which departments were to claim the new diagnostic tool.

  The radiologists were quietly taking up the reins of power. Nuclear medicine—the specialty that images the body with procedures such as PET and SPECT and treats disease with radiation—thought MRI rightly belonged to it. At the Santa Fe meeting of the Experimental NMR Conference in 1986, Walter Robinson, a physician with a background in nuclear medicine, gave a lecture on why MRI should be in nuclear medicine and not in radiology. In nuclear medicine departments, people are familiar with biochemistry and with the kinetics of biological processes, knowledge that would certainly be critical as the field moved forward. And they are familiar with all human imaging techniques, so MRI could become established in a friendly environment. They are familiar with both physiology and physics. Robinson also said, or implied (at least this was the implication attendees took home), that radiologists have no scientific skills and so couldn’t be trusted with the development of the new technology. Furthermore, radiologists have no bedside manner and have to be kept in the back room reading x-rays. The radiologists were not pleased.

  There were other dogs in the fight. Jerry Pohost, the pioneer in cardiac imaging, gave talks in the early years saying that NMR stood for “no more radiologists.” Radiology, on the other hand, with its marvelous membership, money, and power, was not about to let so big a prize get away. The radiologists counterattacked at the Santa Fe meeting, at which the issue of the name was given a full hearing. Since they brought big money to hospitals and could therefore purchase expensive MRI instrumentation, it wasn’t really a fair fight, and the radiologists triumphed. Not all practicing radiologists were especially comfortable; they whimpered about all the new things they would have to learn. In an x-ray or CT scan, black means something solid, white means empty space, and shades of gray can be interpreted according to their density. In MRI, either black or white could be solid, and a lot more besides, depending on how the study was done.

  Leaving Stony Brook

  Paul turned fifty in 1979 and was hoping to expand his laboratory and explore as many ramifications of MRI as possible during his remaining career. He felt stymied in his research at Stony Brook for lack of medical collaborators and lack of facilities. He felt stymied in his marriage, because although the lithium treatments had finally ameliorated Rose Mary’s illness, he had been emotionally separated for so many years he couldn’t find his way back. The two reached a legal separation agreement when the children were in their late teens.

  Commercial MRI systems were being set up in major and minor medical centers all over the world. The equipment was now far, far more sophisticated than any Paul and his students could build in the laboratory. Paul was continually embarrassed by the crudity of his images in comparison to those that could be obtained on commercial machines. In talk after talk he would show his images and have to say that much better ones had been obtained elsewhere. He needed access to equipment that would function at least as well as that available at most medical centers, or he and his work would gently fade away. As important as his work was, no one was about to provide for him anything so costly for pure basic research. He thought it might be possible to share a clinical magnet with diagnosticians, but he had to be free to alter it to develop new imaging techniques. At best, such an arrangement would be complicated.

  He began to search the horizon and was most attracted to the University of Illinois, the main campus, in the farming heartland and surrounded by corn and beans. When SUNY realized their star professor might get away, there was (at least according to the local press) a big retention effort. Paul was to get a $1.7 million facility with three research magnets, including an expensive human-scale system. The newspapers were excited about the size of the offer for research spending and the creation of a new NMR institute for Paul to head, and newspapers both in Illinois and on Long Island played up the large salary (by academic standards) offered by both institutions. But for Paul, salary was not the most important thing. Reporters were surprised when both universities said that the bidding war did not turn on salary but on access to research equipment. “It is not unusual for a top researcher to earn more than $100,000 a year,” reported Long Island Newsday. “One of the key aspects of the negotiations is a guarantee that Lauterbur would have primary control over the NMR devices, sources at both schools said. Many of the scanners, which cost $1.5 million or more, are maintained by hospitals primarily for clinical use. Lauterbur uses the machines for research.”13

  Figure 8.3

&
nbsp; Paul at the controls.

  And there was the rub. Paul saw nothing in the negotiations at SUNY that would give him primary control over the NMR devices. There was a severe disconnect between the publicity being generated and Paul’s experience. David Woods, SUNY spokesperson, addressed the matter this way: “We’ve taken a number of steps during the past year which should provide the basis for developing what could easily become the world’s finest NMR imaging center right here at Stony Brook under the direction of the man who made NMR imaging possible.” Tellingly, David Glass, Stony Brook’s vice provost for research and graduate studies, informed the Long Island Newsday that “[Lauterbur’s] eminence adds to a whole number of departments,”14 and then listed radiology and cardiology as examples.

  Paul’s relations with the medical school, especially radiology, were not good. He thought them untrustworthy; they must have thought him a curious and dangerous beast. Paul felt they didn’t want him fooling around with any MRI system they might get their hands on, and he had reason to be skeptical. His association with the Radiology Department, headed by Mort Meyers, had started well enough, with Paul given a joint appointment and receiving part of his pay from that department. But things soured. Their first collaborative effort involved a young radiologist who brought around autopsy specimens of hearts from children with atrial septal defects. Undergraduate Ed Heidelberger did the imaging.

  The radiologist wrote a paper, with which she won a departmental research prize. Paul and Ed were at her talk, and never once did she mention that she had not done this work entirely alone. Paul banned her from the lab for stealing Ed Heidelberger’s data. Not long after, Paul found a radiology office marked “Magnetic Resonance Imaging”; he had not been told there was such group. Paul took a call from Mort Meyers in which Mort observed that things were not going well, and Paul agreed. They terminated their relationship. Meyers then garnished Paul’s salary to reclaim all of the past support from Radiology funds. The department of chemistry, then headed by Sei Sujishi, found the wherewithal to reimburse the funds. In a final swat, Meyers wrote Paul negative recommendation letters for his job search.