Magnetic Resonance Imaging with Polarized Gases

Stephen Kadlecek. American Scientist. Volume 90, Issue 6. Nov/Dec 2002.

As this article goes to press, my company Amersham Health, is delivering two machines to New York University Medical Center for a study to compare healthy and diseased lungs. The machines are helium-3 (3He) polarizers, and the lungs belong to 200 or so members of the New York City fire department who worked for weeks in the cloud of dust and soot that followed the collapse of the World Trade Center. Some, but not all, of these firefighters now suffer from what is being called the “World Trade Center cough,” although the exact cause and the short- and long-term effects of exposure are not well understood. Georgeann McGuinness, a physician at NYU, is beginning a five-year program to evaluate this condition, and her results might one day improve treatment for all those similarly exposed.

Until recently, methods to assess localized changes or abnormalities in lung structure and function were hard to come by. The two most common pulmonary diagnostic tools in use today are spirometry-the measurement of exhaled air volume and the rate at which it is expelled-and chest x-rays, either conventional ones or those done with CT scans. Helium-3 polarizers, in combination with magnetic resonance imaging (MRI), will add to this short list, allowing microscopic examination of lung structure and localized studies of lung function, which have not been possible before.

The purpose of the polarizers is to align the nuclear magnetization of the atoms in 3He gas. The test subjects will inhale this specially prepared helium, and an MRI scanner will make measurements of where the gas goes and how long it takes to get there. The results obtained with this approach can show extremely detailed features of gas flow in the lung. Similar techniques can be used to reveal how an inhaled gas moves into the circulatory system, to document how blood flows around the heart and to the brain, and to track other physiological phenomena that are difficult or impossible to follow using any other strategy. The medical use of polarized gas is in its infancy, so the extent of future benefits is hard to predict.

Interestingly, the technology that makes 3He polarizers possible was not developed with these applications in mind. Indeed, the story of how polarized gases have entered the world of health care is a nice example of curiosity-driven research in one realm (in this case, atomic, nuclear and high-energy physics) yielding unexpected benefits in another. But to see why polarized-gas technology is now moving from physicist to physician, one must understand a few of the basics of traditional MRI.

Picture This

Most people have seen a standard MRI scanner. Its distinctive tubular geometry, and the cramped quarters it allows for the patient, are determined by the powerful magnet required for operation. Even if you’ve never seen a scanner, you have probably had an occasion to view MRI images and remark on their astounding resolution and clarity, which can take much of the guesswork out of difficult diagnoses. Just as important is the fact that an MRI session has no known side effects (other than perhaps a short bout of claustrophobia). The procedure is completely noninvasive and does not expose the patient to ionizing radiation. The only danger is to people with certain metallic implants, which might heat up or be tugged loose in the strong magnetic field of the machine.

To understand the need for such a strong field requires a brief digression into the magnetic interactions of atoms. As all students of science learn, the elements in the periodic table are arranged according to the total charge of the nucleus, which is the same as the number of electrons in the neutral atom. That classification proves so useful because almost all of chemistry and biology is determined by the interaction of these charges. It is another property of the atom, however, that is critical for MRI: the nuclear magnetic dipole moment. This term refers to the tendency of many atomic nuclei to behave like tiny bar magnets. The dipole moment represents the strength of one of these little magnets, which, like a bar magnet, has both north and south poles. Control over the orientation of these poles is required to align (that is to say, polarize) atomic nuclei.

In most areas of scientific or practical interest–chemistry, electronics and so forth-the interaction of these tiny magnets can be ignored. This is just as well, because magnetic interactions are one step more confusing than electric ones. For instance, the atoms arrayed in the periodic table do not have an orderly progression of nuclear magnetic moments, because the orientation of the magnetic moments of their constituent particles may be such that they add or subtract. In fact, the hydrogen nucleus, which consists of only a single proton, has a greater magnetic moment than most larger nuclei, some with tens or even hundreds of protons and neutrons.

Two features of the magnetic dipole moment of the nucleus are central for MRI, and both can be easily understood by analogy with ordinary bar magnets. In particular, picture a large bar magnet, which represents the MRI machine, placed next to many smaller magnets, representing the tiny dipoles of hydrogen nuclei in the body. First, note that the large field from the big magnet readily polarizes the small magnets (makes them point along the direction of the field). You may have experienced the force that causes this alignment if you have ever tried to bring the like poles of two magnets close together. Not only is there a force pushing them apart, but there is a large torque on each one as it tries to align itself with the magnetic field produced by the other.

The hydrogen nuclei in the body are not able to move very far in response to magnetic forces because they are held in place by much stronger electrostatic forces. But they do respond to the torque, so they tend to line up with the applied field. What happens if one of these protons is pulled out of alignment and then released? Like a bar magnet that is forcibly canted away from the applied field direction and then let go, a jostled hydrogen nucleus oscillates around its equilibrium position. But it doesn’t just shift back and forth; rather, it wobbles like a spinning top. That is, each nucleus remains tipped over and moves in a circle around the preferred orientation.

Physicists refer to this motion (which at first blush appears terribly perplexing, for both tops and nuclei) as precession. As you might imagine, the precession frequency depends on the torque applied to the nucleus. The larger the dipole moment, or the larger the aligning field, the faster the precession takes place.

If you bring a coil of wire near a group of synchronously precessing protons, you’ll find that the magnetic flux through the coil varies with time. As Michael Faraday discovered in the 1830s, such changes in flux induce a voltage across the coil. This voltage is proportional to the amount of flux passing through the coil and to the precession rate. In the case of an MRI pickup coil, the signal is small, but it can be amplified and used to learn something about the precessing nuclei.

Of course, real MRI is more complicated. It requires that the applied field be manipulated to vary in intensity at different points in space, that the precession be made to happen in a carefully prescribed manner and that the waveforms induced in the coil be amplified and processed using elaborate algorithms. The details of how an image is constructed are fascinating in their own right, but they need not be mastered to appreciate the value of polarized gases to medicine. One only needs to understand some simple requirements imposed by basic physics.

First, the nucleus employed should have a large magnetic moment: Stronger north and south poles means more flux through the receiving coil and a bigger signal. A high density of these nuclei is also helpful, because having more little magnets also means more flux. These two general principles explain why the hydrogen nucleus was chosen for traditional MRI. Not only is the proton magnetic moment unusually strong, the human body is largely composed of water (with its two hydrogens) and various hydrocarbon molecules (each with many hydrogens). So a person contains a lot more hydrogen atoms than atoms of any other element.

The second thing needed for MRI is a very powerful magnet. One reason comes directly from Faraday’s law—increasing the aligning field raises the frequency of precession, which boosts the signal picked up in the coil. The other reason is a bit more subtle and requires explanation, because the analogy with oscillating bar magnets that I’ve given is somewhat incomplete-it doesn’t include the random atomic motion that is present at the temperature of the human body. So you must picture, in addition, someone shaking and spinning the small magnets in an effort to randomize their orientation. They will still have a tendency to orient themselves, but the steady-state polarization will depend on the strength of the applied field as compared with the violence of the shaking.

The usual approach for generating intense magnetic fields is to send large currents through superconducting coils cooled in liquid helium to within a few degrees of absolute zero. Despite all the fancy engineering that goes into building such magnets, thermal shaking invariably wins out, and the level of polarization achieved is only on the order of a few parts per million. For the investigation of most parts of the body, this small amount of polarization is sufficient for imaging, but the density of hydrogen nuclei in air is simply too small to register. So airy lung tissue appears pretty much as a void on standard MRI images.

Now You’re Cooking With Gas

Because intrinsic magnetic moments and densities can’t be changed, the problems encountered when imaging lungs can be overcome only by increasing the polarization of the gas they contain. In principle, that feat could be accomplished by boosting the strength of the applied magnetic field, and such an option is being pursued. But the engineering challenges are tremendous, and large gains are unlikely any time in the foreseeable future. One might imagine also decreasing the shaking by lowering the temperature. For instance, cooling lungs with liquid nitrogen (at -196 degrees Celsius) would increase polarization by more than a factor of three-but that expedient is unlikely to be popular with MRI subjects.

The more practical solution is to polarize a gas outside the body, have the subject breathe some in and then make an image before the gas becomes depolarized. Choosing the gas requires some careful consideration of atomic physics and of a property of the nucleus called spin. The spin of a nucleus originates from the fundamental nature and motions of the subatomic particles inside it. It is important to realize that the nuclear spin around a given axis comes in discrete chunks-the corresponding angular momentum must be an integral multiple of Planck’s constant divided by 4pi. All instances of a given type of nucleus have the same spin; they can never change this value by speeding up or slowing down.

As you might guess, nuclei with zero spin have no magnetic dipole moment, making them useless for MRI. This eliminates a good portion of the periodic table from consideration, including such handy nuclei as ordinary oxygen and carbon. In addition, it turns out that only nuclei with the smallest nonzero amount of spin, the so-called spin-1/2 nuclei, are immune to rapid depolarization from large, everpresent electrostatic forces. This leaves a rather select group of candidate atoms (including, among others, helium-3, carbon-13, nitrogen-15, phosphorus-31, xenon-129 and mercury199) from which one must choose on the basis of magnetic moment, chemical binding tendencies and people’s comfort with breathing them. Mercury or another heavy metal, for instance, is probably not a good choice because of its toxicity.

Two candidates that stand out are the noble gases 3He and 129Xe. Because they are chemically inert, these gases are unlikely to produce any long-term adverse effects, when inhaled repeatedly. In the case of helium, this assertion has been demonstrated in countless experiments by college students wishing to sound briefly like Donald Duck. Xenon, although not chemically reactive, is a general anesthetic, similar to nitrous oxide, and has been used as such for many years. In the doses required for imaging, however, the anesthetic properties can be avoided, and exposure is thought to be just as safe as with helium. In addition, the noble gases are attractive because they lack molecular rotation, which, it turns out, often causes an atom to lose its polarization quickly, before a useful image can be acquired.

Before considering the question of how to polarize 3He and 129Xe outside the body, it is interesting to consider where these gases come from. Neither isotope is the most common form of that element, and 3He, in particular, is not found naturally anywhere on Earth. It is only available in appreciable quantities because 3He is the decay byproduct of tritium (3He, a radioactive isotope of hydrogen), which has been produced in bulk since the 1950s for use as a trigger in hydrogen bombs. As nuclear stockpiles age, the tritium these weapons contain needs to be periodically refreshed, and, thankfully, the various nuclear powers save the 3He given off inside their aging armaments. Some governments (but not currently the United States) then sell this rare gas for myriad scientific uses, such as fundamental studies of superfluidity, as a “spin filter” for creating beams of polarized neutrons and for high-energy physics experiments to determine the quark structure of nucleons, just to name a few. Also, because fusion reactions between 3He nuclei do not give off neutrons, this gas is also being seriously considered as a possible “clean” source of fusion energy for the future.

Unfortunately, the supply of 3He is quite limited, and the gas is expensive-currently about $100 per liter, which is about the amount needed to image a person’s lungs. However, technological optimists like to point out that there is an essentially limitless supply nearby: Because of its continuous exposure to the 3He-rich solar wind, the surface of the Moon is full of the stuff. Current estimates, based on measurements of lunar samples obtained during the Apollo project and studies of the solar wind, put the amount of 3He in the top three meters of the lunar soil at about 1013 liters. This is enough for medical imaging, fusion power and just about anything else anybody can think of at the moment.

Acquisition of 129Xe is a bit more prosaic. It is found in the atmosphere at the level of about 1 part in 10 million and can be easily produced by the fractional distillation of air. It would, of course, be tremendously expensive to process ten million liters of air just to take a single lung image, but such quantities are routinely distilled anyway, to produce nitrogen for freezing and packing and to separate out oxygen, used chiefly by the steel industry. The additional expense to pick off the xenon is minimal, and this gas currently sells for about $10 per liter. The cost to separate 129Xe from its more common cousin (131Xe, which is roughly three times more abundant) is probably prohibitive for medical imaging in the short term. So scans taken with xenon will suffer somewhat, being obtained with a gas that contains less of the relevant isotope than it might otherwise, unless the economies of scale push the cost of isotope separation down significantly.

Pumped Up

After the gas has been chosen and acquired, one still needs a method to polarize it outside the body. The most obvious is the brute force approach– exposing the gas to intense magnetic fields and low temperatures, extremes that are unsuitable for live test subjects. This maneuver is theoretically possible, but it would not be very efficient, because even with the strongest fields that can be attained with standard laboratory equipment (about 10 teslas), one would have to lower the temperature of the gas to a few thousandths of a degree above absolute zero to achieve significant polarization. And the cost of the apparatus required to do so would be extremely high.

Instead, my colleagues and I use optical pumping, a technique for transferring polarization from the photons in light to the electrons in atoms. Light, as most freshmen learn in Physics 101, is easy to polarize by reflection, or with simple polarizing filters, and the polarization can be manipulated easily with specialized optics. In 1949, Alfred Kastler demonstrated a technique to transfer the polarization of photons to the lone valence electrons in various alkali atoms (those at the far left of the periodic table-lithium, sodium, potassium, rubidium and cesium). Kastler won the 1966 Nobel prize in physics for this work.

Kastler’s approach was based on the realization that the rate at which an atom absorbs polarized light depends on the relation between the polarization of the incoming photons and that of the atomic electrons. For alkali elements, with just one valence electron and correspondingly simple photon interactions, he could carefully choose the frequency and polarization of the incident light so that the absorption rate was proportional to the extent to which the valence electron was pointing the “wrong,” or unpolarized direction. In addition, investigators later found that it is possible to arrange conditions so that each time an alkali atom absorbs light, the valence electron reorients randomly. Over time, the valence electrons subjected to polarized light become polarized too. And it’s easy to see why: Electrons that are pointing the wrong way are continually pestered by the incoming light to pick another direction, while the ones that are pointing the right way are left alone.

This technique found immediate application in atomic physics research, where it continues to serve today. But alkali atoms are not suitable for human imaging for several reasons—the most obvious being that they all bum ferociously when they come in contact with water or oxygen. However, by the early 1960s, optical pumping methods became available for elements in other columns of the periodic table, and two independent schemes were devised to polarize the inert gases.

Pioneering one of these methods, investigators at Texas Instruments demonstrated that inert gases can be optically pumped if the atoms are excited by collisions with electrons, similar to the excitation process that takes place in a neon light. This route is being studied for medical imaging by investigators at the Institut fur Physik der Johannes Gutenberg Universitat Mainz, although there are technical challenges that make the approach unsuitable for constructing small polarizers that could be used in a hospital or health clinic. For that reason, my coworkers and I have pursued the other approach, known as spin—exchange optical pumping, which is not as difficult to envision as the name might suggest. In essence, atoms of the chosen gas, say 3He, become polarized simply by allowing them to collide with alkali atoms that are themselves kept polarized through optical pumping.

Keeping with the Program

Sounds simple, right? In principle, it is. But many practical problems had to be solved before an efficient gas polarizer could be constructed for physicians’ routine use. The root of the difficulty is that at each step of the polarization process, the gas is being asked to move away from its thermal equilibrium state-that is, to decrease its entropy. Nature, of course, will always find a way to return things to equilibrium, and all one can do is try to slow the process. The challenge is analogous to constructing a refrigerator. In both cases, the laws of physics impose a fundamental limit on the efficiency of any machine, and, in the end, if you turn off the power, entropy will win. But there are engineering choices (for example, adding insulation) that help, once you understand the forces pushing the system back toward equilibrium.

The physics of heat transfer has been understood for quite a long time, so designing a refrigerator is fairly straightforward. But it has taken the concerted effort of many atomic physicists over the last 40 years to understand the equivalent processes governing the transfer of polarization. (At a fundamental level, polarization, like energy, is never destroyed, but it can be “lost” in processes analogous to friction.) And even after all this time, I and other specialists involved in this line of work are still missing several important pieces of the puzzle, which is why spin-exchange optical pumping remains such a vital area for basic and applied research.

In one sense, the difficulties seem surprising. After all, we’re working with an inert gas and an alkali vapor, two elements long prized for their simplicity and theoretical tractability; yet we uncover a tremendous variety of complex behavior. The atoms collide, form a variety of short- and long-lived molecules and clusters, adhere to and diffuse into the container walls, absorb and reradiate light, and exchange electrons with one other. Often, what goes on during these events turns out to be a complete surprise and has important implications for our quest to turn these goings-on into a technology that is medically useful.

Although much basic research still needs to be done, the state of the art is now sufficiently advanced to study the clinical utility of polarized-gas MRI. This process is under way, and the early results look promising.

Medical testing on people began in 1995, when investigators at Duke University produced the first human lung image with a polarized gas. They used 3He because of the inherently clearer signal and because the technology for polarizing this gas was more mature. Within a few years, more than 100 patients had been imaged in this way (primarily at Duke and the University of Virginia), and the sensitivity of polarized gas imaging became evident.

But before this scheme could be made available on a larger scale in the United States, it needed approval from the Food and Drug Administration. The FDA was faced with an unusual question: Is the 3He polarizer a medical device, like a pacemaker, or is the polarized gas a drug, like a medicine for high blood pressure? Each is subject to regulation, but the requirements are very different. In 1998, the FDA ruled that polarized gas is a drug; so, like any other substance given to patients, it must go through a multi-year set of trials with both animals and human volunteers to show both safety and efficacy. (That is, it must be demonstrated that doctors can actually use these images, clear as they obviously are, to help their patients). Trials with 3He are about half way to completion, and doctors are beginning to explore the medical uses of 129Xe as well.

Xenon’s Paradoxes

At first, it may seem that polarized xenon offers no advantages over helium (aside, perhaps, from lower price and greater availability). But it turns out that the two gases behave very differently in the body, despite their chemical similarity. A hint is found in the fact that xenon is a general anesthetic, which says right off that it passes readily through membranes of the lung and dissolves in the blood. Helium, in contrast, isn’t particularly soluble. And unlike many other substances, xenon can get into the brain (many compounds cannot cross what physicians refer to as the blood-brain barrier). Indeed, normal xenon is now administered to monitor cerebral blood flow in stroke patients, using x-ray scans to show the location of this radio-dense element. Interestingly, xenon preferentially collects in fatty tissues, such as the brain and certain types of arterial plaque, which suggests that it might one day be extraordinarily valuable for early diagnosis of many common diseases.

Magnetic resonance imaging with xenon has one obvious advantage over x-ray scanning: It doesn’t subject the patient to potentially dangerous radiation. In addition, polarized xenon has a remarkable ability to reveal something about its chemical surroundings. The property that allows xenon to do this is called chemical shift: The outer electrons shield the xenon nucleus from externally applied magnetic fields, and the degree of shielding changes in response to interactions with the local environment. This phenomenon changes the resonant frequency of nuclear precession slightly and allows one to look at the polarized xenon in an entirely different way-instead of discriminating the xenon by its location, as is done in imaging, one can discriminate the xenon by the type of fluid or tissue in which it dissolves.

The diagnostic potential of xenon is therefore tremendous, even if the images will never be as detailed as with helium. For example, physicians might use the chemical shifts of xenon in different environments to measure the rate at which the gas moves across the lung membranes and into the blood. This number is not just of academic interest: Some pulmonary ailments are characterized by a swelling of the lung membranes or capillary walls, and there is currently no way to gauge this thickening, except at autopsy. So the hope is that by measuring how the chemical shift of polarized xenon changes over the course of a few seconds, physicians will be able to identify people with this condition and find ways to help them.

Similar techniques also show great promise as research and diagnostic tools for neuroscientists. Measurement of the chemical shifts of xenon in the brain show several distinct peaks. Wolfgang Kilian and his colleagues at the Physikalisch-Technische Bundesanstalt in Berlin have tentatively identified two of them with white and gray matter, which has allowed these investigators to study the flow of blood into these two different types of tissue. Work is just beginning, but once the various components of the signal are fully characterized, neurological disease might be diagnosed by looking for abnormalities in these patterns.

Given these many promising applications for xenon, one might consider revisiting the periodic table for other possibilities. No other candidate nucleus has the large magnetic moment of 3He, and it turns out that 129Xe has an unusually large chemical shift, making it ideal for studies aimed at revealing changes in the chemical environment. But the polarizable isotopes of carbon and phosphorus (13C and 31P) also deserve study, because they have moments of acceptable size and are biologically important. Carbon-13, in particular, is chemically indistinguishable from normal carbon (12C) and can therefore be assembled into most of the molecules important for life.

Note, however, that once one ventures away from the inert gases, atoms appear almost exclusively in molecular forms, and many of these molecules depolarize their constituent nuclei almost instantly. A number of possibilities that might otherwise seem attractive based on magnetic moment and biological compatibility, such as sulfur hexafluoride, must be rejected on this basis.

In addition, it appears that optical pumping cannot be used to polarize carbon or phosphorus, although other methods are available and are being pursued (at least for 13C). And neither atom can be administered in elemental form-no one wants to breathe soot, and elemental phosphorus burns spontaneously-so they must therefore be polarized within the molecule of interest. Despite these complications, colleagues in a Swedish division of my company have made such remarkable advances recently that we can foresee the additional steps needed to build commercial polarizers for these elements. This technology is much less mature than the inert gas polarizer, and medical trials have not yet begun, but we suspect that 13C and 31P will eventually find their own special uses, just as 3He is now being enlisted to examine some of the people who became ill after the attack on the World Trade Center.

Perishable Goods

Although my colleagues and I had once toyed with the idea of polarizing helium in a central facility and then sending the magnetized gas on to hospitals as needed, we decided that approach would prove difficult because of the tendency of the gas to depolarize over time. In our lab, for example, the polarized 3He we routinely prepare lasts only a day or two before it needs to be pumped up again. A two-day interval would be adequate with timely deliveries, but it would require that the gas be held in a carefully controlled magnetic field, which might be difficult to arrange in the confines of, say, a FedEx package.

Instead, we elected to construct compact helium polarizers suitable for use in hospitals or clinics. The units built for NYU are about the size of an office desk. Technicians insert at one end a flask filled mostly with 3He plus nitrogen and a little bit of rubidium. The polarizer heats the glass vessel to about 160 degrees Celsius, which causes a small amount of the rubidium to vaporize and mix with the other gases. Circular coils surrounding the hot flask ensure a uniform magnetic field of some 20 gauss, about 40 times the Earth’s background. Laser light, passed through some simple optics to polarize the beam, irradiates this gaseous mixture. The outermost electrons of the rubidium atoms absorb some of the incoming light, polarizing them, and as these atoms collide with the helium, they pass their polarization to the 3He nuclei.

Within hours, most of the helium in the hot flask becomes polarized. After cooling and condensation of the rubidium, the polarized helium is drawn off and put (along with some nitrogen dilutant) in a plastic bag, which will then be given to the patient, who has already been helped into the scanner. After inhaling the contents of the bag, the test subject may be asked to hold his or her breath for a few seconds as technicians work the modified electronics of their scanner to produce what until now they’ve never been able to attain in an MRI machine: a detailed lung image.

With this single study of New York firefighters, one can see glimmers of an exciting future. Doctors will have available to them a suite of new MRI tests, each so safe that it can be applied not only to those who are sick, but also to those who might, because of exposure, genetics or any other reason, be at risk for becoming ill. The earlier physicians can diagnose disease, of course, the more likely they are to be able to help their patients, either by applying a medical therapy or perhaps simply by convincing the patient to stop smoking or to avoid some dangerous environmental exposure. Given the many and varied ways to look with MRI at polarized atoms in the body, the techniques being pioneered now may very well help usher in this new style of preventive medicine.