Paul Lauterbur and the Invention of MRI

by M. Joan Dawson

  Figure 9.1

  Joan Dawson with undergraduate Rachael Tappan.

  I tried hard to make Elise feel that she, not our work, was the center of our world. She was upset when Paul left home for the various meetings, speaking engagements, and other responsibilities that called him out of town. That was often; once I couldn’t remember whether he was in Barcelona or Banff, Alberta. One day as I was driving back from the airport with her sobbing, I tried to comfort her. “Your daddy loves you very much,” said I. Said she: “Daddy loves airplanes, too!” It was the first full sentence she ever spoke. When Paul and I traveled together we always took Elise with us. In the early days I also took a whole separate suitcase of her stuffed animals. It was quite a chore, schlepping those stuffed animals around the world! I would try to arrange a few extra days for a break either in the host city or nearby. Paul would spend these days reading, and I would spend them with Elise. Then he would listen raptly to the stories of her day. We all fell in love with each other all over again on our vacations.

  Oscar the worm is a Lauterbur family bedtime tradition. Paul’s grandfather told the stories to his father, and his father told them to him. Paul told them to each of his three children, and his son Dan continued the tradition with his kids. Oscar has had many and varied experiences. His story always connects with something in the child’s life and suited to the child’s age and personality. Elise, at three, accompanied us to San Francisco and, a child of the Illinois prairie, was frightened of the steep hills. She asked me in great fear as we rode in a cab, “We won’t fall off, will we?” That night Oscar saw steep and frightening hills, and Grandpa Worm had to reassure him that they were safe; they were only anthills. In the great midwestern floods of 1993 Elise helped me fill sand bags. That night in Oscarland it rained and rained and rained, so poor Oscar was flooded out of his hole. But the water went down, and he cleaned up and lived happily ever after.

  The Tale of Three Magnets

  Back in Illinois, Paul had negotiated a suite of NMR and MRI instruments with which he could do most of the experiments necessary to test his ideas. He needed to do chemical NMR spectroscopy to test that the basic physics and chemistry were right and that the idea worked in small samples of cells and tissues, then to test the ideas in small animals, and finally in human subjects and patients. So he needed three systems, the most expensive of which was the human imager. We were to have access to two human imaging machines, one at our local Mercy Hospital and one at the UI–Chicago campus. But best-laid plans oft go astray. For both political and technical reasons, the system in Chicago was never available to us. The human imager belonging to Mercy Hospital had, unfortunately, not made money. We used to joke that these were the only people in the country who had a perfectly good MRI facility and could not make money. We may have been right. After a year or so of financial disaster, a legal dispute arose between Mercy and the manufacturer of the MRI system, Advanced Magnetics. A judge sealed the magnet room and no one had access to it for the next eighteen months. When the suit was finally settled, Mercy sold the magnet for spare parts. Thus did a major foundation of Paul’s research program disintegrate.

  The lawsuit disrupted plans for the small-animal system, too. The magnet was constructed by Varian and placed in shipping containers in Varian’s parking lot, where it remained for eighteen months. Because of the extended lawsuit we were unable to receive the containers. Varian sent us a photograph of the equipment crated and ready for shipment. Paul had it framed.

  The third and least expensive part of Paul’s tripartite plan was a high-resolution spectrometer that was to be placed in my laboratory in the School of Life Sciences. But there was a delay of some months because, while this laboratory had been promised to me, it was not ready when I arrived. This was my personal encounter with the tendency of universities to overpromise space in order to attract faculty. Airlines overbook seats because some passengers will not show up; there is no way to know the exact number of no-shows, and so passengers with valid tickets are sometimes not accommodated. The university, too, played a game of chance in attracting faculty. I had to be satisfied with an old wooden student desk in the corner of a dusty room. I kept my telephone in a fume hood, usually used for storing dangerous chemicals, because there was no place else to put it.

  The long-awaited installment of my magnet in my laboratory was something of an event in the School of Life Sciences. Biologists don’t know much about this NMR stuff, and they were worried about the effect of a high-field magnet on their own research. One day I received a visit from a fellow faculty member whose laboratory was three floors below mine. “Is your magnet messing up my computers?” Since the fringe field of my magnet extended only a few inches below my floor, I was quite surprised. I tried to imagine the characteristics of a magnet that could affect equipment so far away; it would have to be bigger than the laboratory!

  The BMRL Family

  I call it the BMRL family because after the turn of our Biomedical Magnetic Resonance Laboratory on the stage we all, staff and students, felt the closeness of family ties to one another. This was in part due to Paul’s personality, and mine, and that of Debbie McCall, Paul’s administrative assistant. We celebrated Thanksgiving and Christmas together in our home. We celebrated each other’s graduations, birthdays, marriages, and newborns. It was also in part because we went through hard times together, which if they don’t split a family apart make it stronger.

  Figure 9.2

  Debbie McCall. From the personal collection of Debbie McCall.

  Students and postdoctoral fellows began arriving within a few months of each other. Pratik Ghosh, Shachar Frank, Allan Moromoto, Xiaohong Zhou, Doug Morris, and Erik Wiener all began their work in 1987 and 1988. It was fun talking with Paul. Doug Morris said that when you appeared in his office he would throw his pen down, throw up his hands, and inquire “Whatyagot?” And then you got his full attention. Among us were innumerable undergraduate students who got quality research experience almost unheard of in an MRI laboratory. At our peak the BMRL had between forty and fifty affiliated faculty and students, although a very small staff. The enthusiasm was much as it had been at Stony Brook. The students called themselves the “Lauterbur Brain Trust,” and trusted, demanded, of themselves cutting-edge science. Erik Wiener described it as “the Wild West; our past didn’t matter—you were judged only by what you did now.” David McFarlane, an engineer, got the small-animal magnet up and running by late 1989. He had never done anything remotely like it before.

  Ordinarily a professor assigns a student a particular research project. One student, Yihong Yang, loves to tell the story that he waited and waited for Paul to assign him a project, and suddenly realized Paul would never assign him a project—he would have to find one on his own.

  Paul’s students had different things to say about his mentoring, including that they were frustrated by his slowness in editing the papers they wrote for publication. But by and large, they may have in their aggregate hit on most of the complex virtues of the man. Most said he was an inspiration to them. Not just because of his achievements and discoveries, not just because his old gray Mazda was always in the parking lot, but because of his integrity, independence, resilience, and courage. One student offered that Paul was the most ethical man he had ever met. Paul intended them to think for themselves, to think creatively and to distrust received wisdom. He complained that a high degree of education, while necessary for a scientist, could have deleterious effects. “You don’t know you’ve got a really good idea until at least three Nobel laureates tell you it is wrong.” He felt that while you need an intense education, you must “rise above it.”

  Figure 9.3

  The BMRL family.

  Paul had an infinite number of ideas to explore and a rather small number of people to trek around them. “We’re a small laboratory pretending to be a very big laboratory,” my student Estelle Fletcher observed. Everything Paul did was aimed at maximizing the power of MRI, and he, but fe
w others, understood how all the pieces fit together. People in the laboratory were not working on different aspects of the same circumscribed problem, as is usually the case; they were doing research in all of MRI.

  Twenty Years of Productivity

  The work Paul carried out at the University of Illinois was always cutting-edge innovation, an infinity of exotic ideas. He had limited resources; why waste them on something ordinary? “Big companies could devise what I could do with a couple of undergraduates.” So Paul had to stay out in front where they couldn’t get in the way. “There were always plenty of new things to do, rather than refining ideas that had already caught on. And there were plenty of ‘It can’t be done’ statements, so each new idea needed proof of principle.” “The only thing more irritating than the ‘it can’t be done’ crowd was people who saw the wonders and thought anything could be done—no matter if it contravened the laws of physics.” “I couldn’t do GE’s job and they couldn’t do mine.” So Paul approached his science widely and wildly, to the point that some grant reviewers thought he was talking science fiction. It is instructive to look at a few of the MRI projects of his mature years, ages fifty-five to seventy-five, those that succeeded, those that did not, and those that cannot yet be placed in either category.

  Surgical Specimens and Microscopic MRI

  When a pathologist applies his kit of tools to discern disease in a biopsy specimen, there are always some false negatives, when the disease is there but not detected, and false positives, when the disease is apparently detected but not really there. Paul had some hopes, still not realized, that MRI of these biopsies might yield additional information and thus reduce the error rate. For this, MRI must be done on a microscopic level, something that presents unique scientific and technical challenges. For Paul’s early start in this area of research, Allan Johnson, highly respected for his work in MR microscopy, observed, “Paul is responsible for the jobs most of us have today.”1

  While he was still at Stony Brook, Paul and his student Kyle Hedges had achieved a microscopic resolution of 10 μl3, about the size of a single cell and the proof needed to encourage other scientists to enter the field. But it was published only in Kyle’s thesis. Paul’s first publication of the microscopic imaging work in a peer-reviewed journal came out in 1982, and he was still hard at work on the problem at the University of Illinois in 1990. “We are getting to the cells in tissue,” he said in an interview. “The potential advantage of this is we can look inside thick, opaque things and see some of the structural details that cannot be seen by light microscopy.”

  He was expanding the scope of microscopic MRI research in three different ways. The first was to label brain cells from rats with magnetic particles and transplant them into other rats, and then track those cells by MRI. “That tells us where those brain cells go and what happens to that transplanted tissue in the brain of the rat. This is important not only for studying how brains function and are modified and fixed, but also eventually for providing a better basis for transplantation therapy in humans. We’re using these paramagnetic materials in the same way that a microscopist uses stains, to highlight different structures.”2 At the same time he worked with engineers at the University of Illinois to develop magnetic resonance coils less than 1 millimeter long that could be implanted directly into an animal. “We hope to get NMR signals from very much smaller quantities of material than we can now. We will actually put microscopic parts of an NMR machine inside tissues and organs to get a close-up look at things that are happening.”3 Paul was ahead of his time in both these areas of research, and both have much future potential.

  Paul’s most exciting work in this arena was to expand the attainable resolution well beyond that of a light microscope, to the molecular level. One day he came home quite excited and said he had just had his best idea since imaging. We talked about it over dinner. All well and good, but he had actually come up with this idea in 1972 but had not developed it. And he had forgotten about it. He was entering the world where only the electron microscope could previously go, and planning to do so without subjecting the specimen to the special procedures and alterations that are necessary for electron microscopy. The problem he had at least theoretically solved was this: in MRI, the signals received from the sample have a finite lifetime of many milliseconds. During this time the hydrogen from which the signal is received can diffuse up to tens of micrometers, and this is the ultimate resolution of MRI microscopy.

  But Paul thought of a way around this fundamental physics. Instead of looking directly for the macromolecules or cell membranes you want to see, it is possible to find them indirectly by how they restrict the movement of water. From micrometers to nanometers: by this indirect method the ultimate resolution of microscopic MRI becomes 1,000 to 10,000 times better. He called the method DESIRE (one must have an acronym!), for diffusional enhancement of signal intensity and resolution. He titled an unpublished paper “Spin Trek: Voyages through the Magnetic Microscosmos.” DESIRE microscopy has been tried in various laboratories around the world, but is still preliminary. It is now felt that while the technique is sound, advances in equipment are necessary before the technique can be successful.

  Metabolic Imaging and Imaging with Limited Data

  When Paul started MRI back in the early 1970s, a few other scientists, including me, were hard at work learning to find and measure tissue metabolites using NMR spectroscopy, or MRS. (Paul joked that MRS stood for wife, since I was involved in it.) We were very excited because there had never before been a method to watch metabolism noninvasively, and MRS greatly expanded our ability to study and understand cellular function and metabolism. Paul thought it was really cool, and I spoke to him for the very first time when he asked me to explain the results I was showing on a poster at a meeting in Oxford in 1977. He had already shown in 1975 that imaging of chemicals is possible.4 He returned home from the Oxford meeting, I was told, high on combining imaging with MRS, a technique now called metabolic or spectroscopic imaging. This kind of thing quickly caught on, and soon a number of scientists were trying to do it. Our meetings on the topic were very exciting in those days, as we got our first peeks at the achievements of other groups.

  But we had one big problem: spectroscopic imaging just wasn’t sensitive enough, and the region of localization of the metabolites required too many encodings to be practically useful. Leave it to Paul to come up with the answer! To calculate the image of metabolites, he said, “Let’s use a priori information to reduce the amount of metabolic data needed to that which we are able to obtain.” Look at it this way: if you have a blank screen you need to test the whole screen to determine where the data points came from. But if you already have an image—of the head, for example, obtained by conventional MRI or anything else—you can constrain the calculation. For example, you know that none of that metabolic data came from outside the head image, or in the skull. This allows you to localize metabolic signals in high resolution from a small set of spatially encoded measurements, taking advantage of something you know independently from that data set. He called this method SLIM (signal localization by imaging).

  Paul believed SLIM played a special role in our joint research; it was a vehicle for the intersection of my own interests in metabolism with Paul’s in imaging. It is almost the only area in which we published joint papers. Some people were intrigued by the SLIM idea and others horrified when it was first presented. “It violates the Heisenberg uncertainty principle,” some said. As a result, Paul’s SLIM grant applications were declined multiple times. But Paul would not give up an idea simply because other people objected. He believed a priori information should be incorporated into the formulation of an imaging problem to significantly reduce the number of measurements needed by the conventional imaging methods. Paul later said he had always felt that SLIM was important not just for a particular application but for broadening ideas of how imaging should be done.

  Together with Zhi-Pei Liang, who was first a postdoc
in Paul’s laboratory and later a professor in the Department of Electronic and Computational Engineering at the University of Ilinois, Paul set out to develop the original SLIM technique, ignoring the shouts of disparagement from around our little world. The first version of SLIM was very rudimentary and did not show as much enhancement of image resolution as we had hoped under practical imaging conditions. But Zhi-Pei continued to work on the mathematics of the image formation, and things got better and better. I tended to think of Zhi-Pei as the leader in this project because Paul often complained that he himself wasn’t a good mathematician. Zhi-Pei contradicted him.

  One day he and Paul realized that a priori information could be used to improve dynamic imaging as well. They hashed it out in one of their usual evening conversations, when they could talk without interruption. This time Paul came home quite late. I was already in bed, and I sleepily listened to this exciting new idea. “Hey,” he said, “maybe we could image the beating heart—now just how useful it that?” The idea was to obtain a high-resolution image of the heart at one point in its beating cycle, as was now being done in a clinical setting. That image could then be used to constrain the calculation of the images of the heart as it changes throughout the heartbeat. Before, we had thought we needed to calculate separate images for each point in the heartbeat, but now we would calculate only the change in the shape of the heart as it goes through its cycle. This new method, which they named RIGR (reduced-encoding MR imaging with generalized-series reconstruction), needs far less data and can be done much more rapidly. Extending RIGR further, they developed DIME (dynamic imaging by model estimation), which enables real-time cardiac images to be obtained with very limited, undersampled data. The strongest SLIM opponent would be impressed by the results. Constrained imaging with limited data is now a hot topic and well accepted by the imaging community. Paul would have been happy to see that his pioneering efforts and ideas in constrained imaging with limited data are finally making practical medical impacts.