Those Sure Are Tiny Signals!
Although there was still much work to be done on the mathematics, Paul had essentially addressed the first problem of MRI: yes, you can make an image from NMR signals. Now on to question two: Would there be enough water proton signal to image large objects in low magnetic fields, and provide useful resolution in practical acquisition times?
Paul’s own words:
I then asked myself “Could you ever get a big enough signal from something as large as a human being, for example?” That was not obvious because the small fraction of an inch was the usual size of the specimens that people were using. But could you do it from something two or three feet in diameter? And so I did some calculations that were standard textbook and figured that it was indeed possible—barely, but it was possible. Those simple calculations revealed that physics allowed an adequate signal-to-noise ratio.
I know some other people (John Waugh, an important innovator in NMR, has said this about himself) at the time—I heard about this later—who had not done such calculations, but when they heard about this work, they said they knew there was not enough signal, they didn’t even have to do the calculations. You never want to know too much, and in particular you want to be a little bit skeptical of what you think you know.8
Can Big Enough Magnets Be Built?
Having settled the first two criteria for establishing the feasibility of MRI, Paul moved on to the question of whether anybody could build magnets big enough to put people in. Chemists were typically using test tubes of 2 mm diameter; the large size was 5 mm. To go from this to the size of the human body boggled many a mind. Among the early objections of NMR spectroscopists to the concept of MRI was that even if it worked in principle, it would never be possible in practice because magnets could never be built big and strong enough, and uniform enough, for human imaging. But spectroscopists don’t know that much about building magnets for other purposes. In the literature unrelated to NMR, Paul found,
There were examples of fairly large resistive magnets with fairly uniform fields . . . but more importantly there were also papers describing how to design them. It looked quite easy (and was), so almost everything had fallen into place to justify my instant faith that I had stumbled across an idea that was worth spending most of my time on for the foreseeable future.9
From the beginning, Paul considered magnet building to be a matter of simple engineering, with no new scientific breakthroughs required. They are “mere gadgets,” he said. “If physics said it could be done, engineers could do it.” The many engineers who, over the years, made clinical MRI possible, and who continue to perfect the magnets and make them more versatile, may not always think of it as simple, but development proceeded as Paul foresaw.
It is interesting that today, many people believe that somehow research on superconductivity (loss of electrical resistance at very, very low temperatures) gave rise to MRI. Superconductivity allows magnets to be built for higher fields, and clinical MRI is generally done using superconducting magnets. But the first MRI systems used electromagnets with a field strength of about 0.1 Tesla, and there was real concern that “supercons”—going to 0.5 Tesla and higher (now 9.4 T for some MRI systems)—would introduce problems that could not be solved. We entered the field strength wars. Our meetings raged with arguments, sometimes quite personal, over what the optimum field strength for clinical MRI would be. Some people shouted about field-dependent errors, and others about the need for big magnets to yield the largest possible signal. Personal insults flew around the meeting halls.
Figure 6.1
Paul’s 1972 drawing of what a clinical MR magnet would look like.
Figure 6.2
Paul’s 1972 drawing of what a clinical MRI suite would look like.
Figure 6.3
Paul’s 1972 drawing of what a control system for a clinical MRI system would look like.
How to See Contrast When It Isn’t Really There
Paul understood and recorded in “The Notebook” that dynamic behaviors of atomic nuclei could be incorporated into an NMR image to enhance the contrast between different parts of the body and between normal and diseased tissues. In the first MR image that Paul published, the variations in amount or intensity of water protons from one region to another supplied the image contrast, and in the second image, published in the same paper, he showed that the rate of signal decay could provide additional contrast.
There are now plenty of ways to enhance contrast, and in clinical images the contrast is usually provided by a combination of intensity and the dynamic behavior of the signal. The proton signals from tissue water, for example, have different relaxation times in different organs, and these are sensitive to physiology and pathology. An image that discriminates between higher and lower concentrations of protons and also between quickly and slowly relaxing protons is clearer than one based on concentration alone.
Over the next thirty-plus years, many useful new ways of generating physiology-based image contrast have been achieved, including contrast due to oxygen utilization (known as functional imaging), metabolite concentrations (spectroscopic imaging), blood flow (MR angiography) and water diffusion within a tissue (diffusion and diffusion tensor imaging), which produces great pictures of nerve fiber tracts. Because the signal comes from the atomic nucleus, the MR signals can produce a wide variety of important information. Paul had predicted the rudiments of all of these actual developments on the night at a Big Boy Restaurant in 1971.
Today, when patients undergo MRI, they often are injected with a material that makes the organ or tissue of interest show up more readily on an image. This material is called a contrast agent and is specially formulated to heighten the contrast of an image and improve diagnosis. MRI contrast agents are comparable to the density contrast media used in x-ray imaging or to the radioactive tracers used to obtain PET images. Paul was an early champion of the use of contrast agents, when others thought they could have no value. The controversy was often quite heated. He introduced the concept of paramagnetic contrast agents in a paper coauthored with Helena Mendonça-Dias and Andrew M. Rudin in 1978.10 That same year he introduced the idea of using chelating agents (binding agents) to make the contrast material more effective. Both innovations were met with great resistance. He also “had a leading role in the recognition of the potential value of using ferromagnetic materials as contrast agents for NMR,” according to Hal Swartz, a leader in this field.11
Eric Wiener, a former postdoctoral fellow of Paul’s and an expert in contrast agent research, recently summed up Paul’s contribution: Paul did the first studies of paramagnetic contrast agents and the first studies of paramagnetic chelates, and he had visions of inventing a set of in vivo magnetic stains like those used in optical microscopy, and that means targeting agents. In the 1980s, when Paul was trying to introduce the idea of magnetic contrast agents, people were saying, Why do you need to improve when you get such good images? Why do you need to inject (possibly noxious substances) when you have this beautiful noninvasive technique? You are making it invasive, negating the very thing that it is so good at. Well, today almost every imaging study in medicine is done with contrast material. So he did two important things for the field. First, he pushed the idea when no one thought it was necessary, and second, he developed the basic foundation from an NMR perspective. Those papers with Mendonça-Dias were seminal papers. Even if the agent you first use is toxic, it is proof of principle, and you can detoxify it.
And so the contrast agent wars began. Those in favor were sure that contrast agents could help in diagnosis and those agin’ thought increasing field strength would make contrast agents unnecessary. We were in quite a muddle for a while.
Pharmaceutical companies began to see the light in about 1982, and MR contrast agents eventually became a billion-dollar industry. Inevitably, legal disputes arose about who invented them, and these are not settled even today. Although Paul invented no specific contrast material in use clinically, he
was the first person to think of paramagnetic and ferromagnetic contrast agents and of the use of binding agents (chelators) to increase the amount of contrast agent that could be safely delivered. These ideas are in notes he made even before the first paper on MRI came out in Nature and in a grant application to the National Institutes of Health in 1977.12 All this was well before any commercial interest. Why did he not patent? He thought these concepts were too simple and obvious.
Imaging in Multiple Dimensions
Paul also realized, and recorded in “The Notebook” on that fateful September night in 1971, that MRI would apply to multidimensional images. Applying the magnetic field gradients in two dimensions produces two-dimensional images; three-dimensional images are produced by applying gradients in three dimensions, and so on. A fourth dimension could be used for looking for variations in time, and a fifth for the rate of change in variations in time. Or a fourth could be used to identify specific biochemicals and a fifth for their rates of change. The amount of signal available sets practical limits on the number of dimensions that can be used.
From the beginning, Paul emphasized multidimensional imaging, and he believed three-dimensional imaging would produce the most useful results.
All of the previous uses of magnetic field gradients [were] in a one-dimensional way, very specific to the [particular] experiment and with no intimation of generality in the process. But from the very beginning I was thinking of it as a general process, actually. Although the citations often refer to two-dimensional [imaging], that was a more practical way than the natural three-dimensional procedure, which was not as easy to implement both technically and mathematically. [Two-dimensional imaging] was really a stopgap solution.13
He was plagued, as in so many other things, by assertions that three-dimensional imaging could not be done. Those who asserted this, Paul said, were importing into MRI artifacts that were relevant to other forms of imaging.
On February 12, 1972, only five months after he conceptualized MRI, Paul wrote out on white lined paper a possible sequence of “magnetographic” studies in the diagnosis of a brain tumor, and sketched what these images would look like. (This was the only time he used the term “magnetography” because the name was already in use for instruments used to map the magnetic fields of the Earth or Solar System.) His sketch showed three dimensions.14
Paul showed his first actual three-dimensional image at a meeting in May 1973, on a coconut that still exists.15 In the early 1980s, he demonstrated true three-dimensional imaging of limbs,16 synchronized three-dimensional imaging of the beating heart and three-dimensional imaging of the head,17 and was at work on three-dimensional image display18 similar to that available on PCs many years later. In the late 1980s he displayed three-dimensional electron spin resonance imaging. For three decades, three-dimensional imaging was not routinely performed, largely because the rather stupid computers on clinical MRI systems couldn’t handle all the information. Stupid computers in turn were largely used because classically trained physicians were used to thinking about two-dimensional images and did not demand three-dimensional capability. Radioimaging was always done in two dimensions, and CT scanning was always done in two-dimensional slices. For these reasons, MRI took off in a direction that Paul had not anticipated and did not want to go. He thought two-dimensional imaging appropriate for Edwin Abbott’s Flat Land, in which the world is imagined to be two-dimensional, but not for the solid physicality of ourselves.
In the beginning, Paul was also trying to use color to help in the visualization of MR images, but doctors wanted nothing to do with this, either. At last, younger physicians and scientists, born after MRI and unencumbered by the past, demanded true three-dimensional imaging and color coding. At the more advanced imaging centers, diagnosticians work from computer graphics in which the diseased tissue is highlighted in three-dimensional space, and can be sliced and diced in any desirable way. Additional dimensions are used for showing the concentration of specific biochemicals, the rate of metabolism, or the rate of blood flow or oxygen utilization in different regions. All of these seemed novel and modern when (in the 1990s and later) they were first accomplished, but Paul had anticipated them all from the beginning.
Patenting
The question of a patent came up almost immediately. Paul turned to a friend and fellow consultant on the board of NMR Specialties, Edward Welsh, a patent lawyer based in Pittsburgh. (To describe Ed to me, Paul quoted him: “I love litigation. It’s like stuffing a wet noodle up a wild cat’s ass.”) They worked seriously on the project but had a falling out over an ethics issue that had nothing to do with the patent application, and dissolved their partnership.
Paul then turned to the university.19 The evaluators at the Research Corporation, which handled patent matters for SUNY, were confused about the concept Paul was describing and thought it was the same as CT, then a brand-new technology. Paul tried to help them over this misunderstanding, but to no avail. The final letter of denial, dated February 20, 1974, explained their “inability to identify a potential market of sufficient size to justify our undertaking the patenting and licensing of this invention.”
Paul then applied to the university for permission to self-patent, but never received a response. A patent must be applied for within one year from the date of publicly revealing an idea. Paul was busy and impatient. He decided to publish, but didn’t manage to take out a patent within the following year. To him the patent issue was of secondary importance, and he just didn’t get around to it. He was always more interested in spreading the idea as widely as possible than in patenting it. This attitude cost him a personal fortune—Paul later told a reporter, “But for that decision we might be holding this interview on my yacht off the Riviera!”20 Though he sometimes rued the price, he never regretted the decision. Francis Bonner, Paul’s department head, was infuriated by the whole mess Paul’s “nonpatentable idea” revealed and spearheaded an effort to put better patent procedures in place.
With issues of the basic patent decided (there would be none), the whole scientific community was free to explore the implications and applications of MRI. Thus the field developed much more quickly than it would have if the concept had been proprietary. Paul energetically took his ideas to colloquia, conferences, and seminars around the United States and the world; he practically lived out of his suitcase. He has often been praised for his helpfulness to others trying to enter the field. This is mostly the result of his very open nature; he believed strongly in the free exchange of ideas that is so important to scientific progress, but it is also due to his failings as an entrepreneur. Paul tried, but never became wealthy from MRI. Instead the technology progressed and became clinically useful, and then essential, at an extraordinary rate.
The Nature Article
“It took a rather impassioned argument and rebuttal to get that first paper published,” Paul reminisced over a decade later.21 He worked alone on his ideas during the busy academic year of 1971–72. All of the earliest theoretical work was done by hand in pencil on a yellow legal pad. Paul’s favorite time for this, and his only time, was during seminars given by his colleagues or invited speakers. Paul always found information transfer during lectures and seminars far too slow, so he split his mind between listening and thinking about his own projects. He joked about how the speakers at the time must have thought he was very involved in their talks, pondering and writing notes all the while, when he was really trying to figure out how to treat the imaging data mathematically. It was always so. He sometimes annoyed me while watching an opera or newscast on television by interrupting my more studious attention with comments about scientific work.
Paul had found early that by using only three linear field gradients, he could obtain the information needed to reconstruct images of two test tubes by projection reconstruction. But this was all that his equipment and developing methodology could support at the time; an early image of three tubes consisted of three bumps in the right location, but no
t a stunning demonstration of principle. Years later he laughingly explained the differences between the first successful experiment and modern MRI. “The two blobs that represented the tubes showed up. That was proof of principle.”22 The two-test-tube model did not bring to anyone’s mind modern MRI.
The first paper on MRI, “Image Formation by Induced Local Interactions: Examples Employing Nuclear Magnetic Resonance,” was only five paragraphs long.23 While the technical and experimental thrust of the work was MRI, Paul emphasized zeugmatography, the name he had invented for his technique as a new principle of imaging, which could be applied to many other kinds of imaging as well. He ended the paper with a paragraph explaining many possible variations on the technique, including the fact that “analogous experiments in other regions of the spectrum should also be possible.” In between he squeezed images of his two test tubes, with and without relaxation contrast.
The paper was submitted to the prestigious British scientific journal Nature, and promptly turned down in a form letter. The paper introducing MRI thus joined a long and honorable list of outstanding scientific achievements the first presentations of which were turned down by Nature or its American counterpart, Science. According to Paul, “You could write the entire history of science in the last fifty years in terms of papers rejected by Science or Nature.”
The principle of MRI was so different from any other way of making images that, in its glorious simplicity, it was originally difficult to grasp. Nor could people see any usefulness to the concept. The basic idea and the simple experiments were initially regarded as trivial, unpatentable, even unpublishable.
Paul appealed in a letter that was longer than the paper itself, resubmitting it with passion and more precise explanation. The letter read, in part,
Paul Lauterbur and the Invention of MRI Page 11