The Berkeley Science Review: Articles

A Magnet With a View
Exploring new frontiers in MRI (view PDF)
by Natasha Keith

The MRI scanner, now a common piece of equipment in most hospitals, is many things to many people. Doctors see it as a lifesaving diagnostic device; patients might see it as an enormous, noisy tunnel of frightening imposed immobility. To researchers at UC Berkeley, the MRI scanner is an expression of science at its most beautiful, and several groups are working to diversify the kinds of problems MRI can handle, starting at the roots.

The average person may wonder how a technique as well-established and commonplace as MRI could be radically expanded. MRI is a trusted imaging technique that already gives us high-resolution images of any part of the body. Aside from the usual tinkering for bettering resolution and making the technique cheaper to implement, what more can there be to study about MRI?

The answer, it seems, is a lot. From physicists to chemists to bioengineers, UC Berkeley researchers are working to revolutionize the field.

How MRI works
What we now know as MRI (Magnetic Resonance Imaging) used to be called Nuclear Magnetic Resonance (NMR), but the term "nuclear" caused concern among patients, so the name was changed to MRI for medical applications. These days the two terms still exist to describe two different expressions of the same technique. MRI refers to the variant that produces two-dimensional images of large objects or animals, whereas NMR refers to the nitty-gritty imageless version that is used to study molecules.

While the fields of biochemistry and medicine tend to deal with interactions between molecules, MRI is based on the physics of the atoms themselves. This makes MRI unique, both in its theory and its high level of safety. Whereas diagnostic tools like X-rays and CAT scans gather images by pelting the patient with high-frequency radiation that scatters off tissues and can damage them by breaking chemical bonds or creating free radicals, MRI functions by observing slight changes deep in the center of atoms upon gentle probing with low-energy, radio-frequency rays. MRI doesn't bother the chemistry of the patient—proteins, biochemical pathways, and molecules are all left intact and untouched. Instead, information is collected from the whispers of the atoms' buried nuclei.

To make those whispers audible, a few physical tricks are necessary. The fundamental principle, which was discovered in the 1930s, is that nuclei of certain atoms have a property that physicists called "spin"; in most cases relevant to MRI, this makes spins act like little magnets that can be pointing either up or down. In a normal distribution of atoms, there is no advantage for the spins to be in one state or the other—both are equally probable. But in the presence of a magnetic field, a tiny majority of the spins align with the magnetic field, causing the entire sample, such as the full body of a patient in an MRI machine, to spontaneously act like a weak magnet.

This overall magnetism can then be manipulated with short pulses of radio waves. The low-energy radiation is the same frequency that is picked up by an FM radio, but instead of being tuned to a radio station, it's tuned to a particular kind of nucleus (for example, carbon or hydrogen). For a brief moment, a burst of these radio waves pokes the magnetization, causing it to change direction. When the radio wave burst ends, the tipped magnetization starts to swirl in a spiral as it drifts back, or relaxes, to its original in-line-with-the-magnet orientation, just like a child's spinning top when tipped off its axis. This spiraling relaxation causes the overall magnet-quality of the patient to drift in different directions, a phenomenon that can be measured by very sensitive instruments. Not a molecule has moved due to the magnetic mayhem; only the curious "spin" property of the atomic nuclei participates in the dance.

Poking the nuclei and watching them relax is just the beginning. When physicists first discovered they could do this, they were bewildered to find that not all nuclei behaved the same way. In a chemical sample that contains hydrogens bonded to different partners, each kind of hydrogen responds to the poke differently, giving off a slightly different signal as it relaxes. Scientists eventually figured out that atoms participating in different kinds of chemical bonds felt slightly different magnetic fields due to the electromagnetic properties of their neighbors. This excited the chemists, because it meant that NMR could be used to explore the chemical properties of a substance. Instead of being trapped in the center of the nucleus, the information in the NMR signal also said something about an atom's surroundings.

They later learned that by using "pulse sequence" schemes to deliver elaborate bursts of radio-frequency radiation to the sample at precisely timed intervals, they could do a great deal more than just see a unique signal for each atom. They could actually emphasize or silence complex signals based on their needs, much as an orchestra conductor can bring out the voice of a single instrument while quieting the rest of the ensemble. It was a new way to see the invisible.

Some years later, a form of NMR entered the medical establishment when researchers developed a new application focused on the signal of water (the main component of most animal tissues) that could generate a three-dimensional image of the data. Since most animal tissues are made up of different quantities of water, the NMR signal from water in a patient produces different levels of brightness in different organs, enabling doctors to see blood vessels, fatty tissues, and bone. MRI was born.

Overall, MRI requires four things: an exquisitely tuned magnet with a very straight magnetic field (donut-shaped works best, with the magnetic field flowing through the center), a machine to produce radiowaves at appropriate intervals, a detector to read the signals from the patient, and the programming know-how to turn this data into an image. With ingenious manipulation of the pulse sequences, an MRI technologist can perform hundreds of different experiments that coax out different forms of information. Some experiments highlight the water-rich spinal fluid in the brain, some hone in on tumors, others concentrate on regions where liquid is moving to find blood clots or hemorrhages.

In spite of its utility, MRI still faces some problems. One issue is the expensive upkeep of the large and extremely precise magnets, the strongest of which use delicate electronics cooled by liquid helium and are tended to by numerous technicians. Since the magnets can interfere with distant electronics and can be outright lethal to people with pacemakers or magnetic health aids, most MRI scans must be done in a separate MRI facility with highly trained technicians. A few famously gruesome stories exist of inexperienced technicians bringing metal objects, such as wrenches, near a functioning magnet, resulting in a deadly blow to the patient inside.

A second issue is claustrophobia. Patients getting full-body or head scans must lie immobile at the center of a cramped, roaring tunnel in the magnet for nearly an hour, which can be traumatic, and is nearly impossible for children (who are usually put under general anesthesia for the procedure). Moreover, MRI is not useful for scanning organs that are in constant motion such as the heart or lungs, or imaging the smaller blood vessels of the vascular system. Images become distorted by metal implants, and it is not possible to unambiguously detect many kinds of tumors.

There is a booming and boisterous community of researchers across the world that work to make MRI more effective, more affordable, and more comfortable. But at UC Berkeley, without the presence of a medical school or hospital beds, researchers in the Clarke, Conolly, Wemmer, and Pines labs are taking a step back to dream up deeper and wilder ways to revolutionize MRI.

Enough with the fussy magnets!
The Pines, Clarke, and Conolly labs have been working on ways to make MRI cheaper and more accessible. Instead of tweaking the established MRI technology, they're going back to basics, and finding ways to rethink the equipment. One of the most frustrating pieces of equipment in a standard MRI laboratory is the magnet. Given the expense of the standard MRI magnets, what would it take to do MRI with smaller magnets, one-sided magnets, or the magnetic field of the Earth alone?

One idea from the Pines and Clark labs would improve the cost, accessibility and comfort of MRI all at once: how about doing MRI with smaller magnets, one-sided magnets, or no magnet at all? If the medical establishment could exchange their helium-cooled superconducting magnets (one million dollars plus operating costs) for small, permanent, hand-held magnets ($50) or the Earth's permanent magnetic field (free), MRI would indeed experience a revolution.

In the case of the smaller or one-sided magnets, there are two main challenges: magnetic field strength, and magnetic field uniformity. At low magnetic field strength, the signal strength is compromised, since fewer spins align with the magnet. With one-sided magnets, the magnetic field is wiggly and irregular, which causes every nucleus in a sample to act differently based on the magnetic field it happens to receive, and makes data analysis nearly impossible.

To get around the magnet problem, all three labs are using a technique that is called "Prepolarized MRI" (or PMRI). It builds on the idea that you don't actually need one very strong and homogeneous magnet to do MRI. Rather, you can use two smaller magnets: one that's strong and inhomogeneous to align the spins momentarily, and a second magnet that is weak and homogeneous to perform the imaging.
The Pines and Clarke groups are trying to push the strength of the second magnet as low as possible. The key appears to be mixing PMRI with carefully derived pulse sequences, which manipulate the spins to make it look like the field is homogeneous.

This innovation takes care of how signal is produced, but in order to read signal from such a weak magnet, it's also important to have a very sensitive detector. The Clarke group has been developing a new kind of detector, called "SQUID" for Superconducting Quantum Interference Device (see BSR Spring 2004). Using a combination of physics and engineering, the group has successfully built niobium and aluminum oxide detectors that can image at extremely weak magnetic fields. The SQUID detector has no preference for signals from strong or weak magnetic fields; it can detect the full range of frequencies equally well.

What if you take away the detecting magnet altogether? Not using a magnet means relying on the magnetic field of the Earth, which turns out to be gorgeously homogeneous but extremely weak. Louis Bouchard, a post-doctoral researcher in the Pines lab, is enthusiastic: "There's new physics coming out about this; it turns out the Earth's field is so homogeneous, we can see signals with no magnet that we've never been able to measure before with our largest magnets!"

Together in the fight against fussy magnets, the Clarke and Pines labs have already had some victories. In the case of using small and one-sided magnets, their pulse sequences and SQUIDs are already working, creating small, low-resolution images. The Clarke group has found that it is even possible to image through metal with a weak magnet, and the Pines lab has produced simple MRI images of objects using only the magnetism of the Earth for their readout magnet. Creating sufficiently high resolution images for medical use from these small-magnet or magnetless systems will be the hard part, but the first steps have already been taken.

Copper wire and petite magnets
Steven Conolly, from the Department of Bioengineering, is also interested in taking the pressure off the costly magnets, but he takes a more pragmatic view. "If you go to an MRI conference for researchers, it's full of all this cool stuff about SQUIDs and fancy tricks with big magnets, but if you go to the button-down radiology conference, they think that's all just hippie craziness," Conolly says. But he believes that the cost of doing MRI can be drastically lowered without necessarily using SQUIDs or throwing out the magnets altogether.

Researchers in the Conolly lab also use the PMRI setup, but instead of trying to use ultra-low magnetic fields, they are working with intermediate-strength readout magnets that are cheap to build and portable. By compromising between magnet strength and signal, Conolly's MRI technology doesn't require the high-level SQUID detectors, but rather uses inexpensive detectors made using a special kind of woven copper wire, called Litz wire. Litz wire, which was invented in 1910, can carry very high electric current without experiencing high resistance, meaning these magnets can create strong magnetic fields without expensive supercooled electronics.

With prepolarization, Litz wire, and a great deal of clever electrical engineering, the Conolly group has built a wrist-scanner and knee scanner that each cost only about $65,000. These have already produced MRI images that are comparable to clinical diagnostic MRI images. "At present, many researchers in the US don't seem to think there's a need for cheaper ways of doing MRI," Conolly says. But he points to the rising insurance costs and the fact that supercooling liquid helium is a limited resource, and, most practically, the move against unnecessary expenses as rationale for building a less expensive MRI. "In current clinical practice, ultra-sound is used eight times more often than MRI, in part because it's inexpensive, quick, and portable. Cost does matter."

Radiologists and orthopedic surgeons in the medical profession get especially excited when they learn that the prepolarized MRI setup has another major advantage: it enables high-resolution imaging around metal, which is impossible with conventional MRI. Metal implants are slightly paramagnetic, meaning that the metal produces a distortion of the magnetic field that creates artifacts in the MRI image that can look like bright spots or black splotches that obscure the region around the metal. "When people have a metal implant, they sometimes experience pain near the implant, such as in bone reconstruction or metal pins in the spine or joints. It's a major drawback of standard MRI that you can't see those areas," says Conolly. Prepolarized MRI is nearly immune to this problem, and the Conolly group has successfully shown that metal pins in a forearm and a knee can be imaged with their homemade magnets for one tenth of the cost of a typical MRI scanner.

Clever contrast agents
Although the paramagnetism of metals often interferes with the MRI image, the paramagnetism of liquid compounds can be put to good use. One commonly used technique in clinical MRI is to inject a liquid paramagnetic substance, called a contrast agent, into the patient, and monitor the changes in the MRI image. Wherever the contrast agent goes, the MRI signal is either enhanced or reduced. Doctors can use this to highlight specific features of an image, such as blood vessels.

One contrast agent commonly used in hospitals today is gadolinium-DTPA (or Gd-DTPA), which is routinely used when looking at the blood vessel networks of tumors. But Gd-DTPA leaks through blood vessels, making imaging more difficult. Also, a large amount of contrast agent must be added to image successfully, and Gd-DTPA causes serious health risks for patients with certain kidney disorders. Louis Bouchard, a postdoctoral fellow in the Pines lab, would like to develop a new kind of contrast agent that has better sensitivity, better dispersal in the vasculature, and guarantees fewer side effects.

The difficulty is that the paramagnetic substances that are best for MRI tend to be highly toxic. The Pines group is now investigating using one such MRI-happy element, cobalt, and avoiding the toxicity problem by embedding it in envelopes of an element that is known to be safe in biological systems: gold. With these tiny wontons of gold-wrapped cobalt, the patient would be shielded from the toxic effects of the cobalt while reaping the benefits of better imaging. The nanoparticles, which are 50 nanometers in length, could be administered in extraordinarily small doses and would stay suspended in the bloodstream for a few hours, enabling the difficult and desirable task of imaging coronary arteries.

A group of researchers from the Wemmer and Pines groups is interested in using contrast agents to follow the movement of individual cells or molecules in the body. Currently, their efforts involve xenon, an MRI-active nucleus that can be induced to give off a strong MRI signal on command. A noble gas, it generally doesn't interact with any other elements, and as a result, it is almost never found in living organisms. But the Wemmer and Pines lab have found, in collaboration with synthetic chemists in the Frechet group, that xenon can be "trapped" inside a molecular cage made of organic molecules. This cage can then be attached to a long molecular tether, which can be engineered to bind tightly to certain kinds of proteins, certain kinds of cells, or even certain kinds of microorganisms. This combination of caged xenon and tether then acts like a bright flag, or biosensor, for the molecules of choice.

Although xenon imaging has been tried in humans (by having the patient breathe xenon gas), the biosensor technology has not yet been tested in living systems. Researchers in the Wemmer and Pines lab are taking that next step now. Monica Smith, a graduate student in Dave Wemmer's lab, is trying to create a tether that will bind to receptors on the cell surface, to see if the cells can effectively internalize them. "Once we find an effective way to attach the biosensor to any kind of target," Monica says, "the options for what you can image with the biosensor are seemingly limitless." Tyler Meldrum, a graduate student in the Pines lab, also works on the biosensor project, but instead of attaching the biosensor to a cell, he is trying to slip it into a virus. If the xenon biosensor passes these biological tests, it could be used for a number of diverse applications.

Meanwhile, the Conolly lab is attempting to follow the movement of individual living cells using a technique called Magnetic Particle Imaging, or MPI. In MPI, magnetically-active tags are attached to molecules of interest and then followed by applying a specific pattern of magnetic fields to detect the magnetic particles.

One application of this technique is in monitoring stem cells, the immature, rapidly replicating cells that are able to differentiate into many possible mature cell types. The dream of stem cell therapy is to inject these cells into a patient to heal tissue damage. For example, in the aftermath of a heart attack, a simple injection of stem cells could start an immediate rebuilding of the heart tissue without surgery.

However, it has been difficult to monitor the movement of these injected cells. "Following stem cells is a combinatorial problem," Conolly says. "If each cell can be transported through 20 different pathways, can sit on 20 different kinds of protein scaffolds, and can turn into one of 20 different kinds of cells, there may be 8000 possibilities. But if we stick a magnetically-active tag on the stem cell, we can follow an individual stem cell from start to finish." In this form of imaging, there would be no anatomical image of the patient, just little magnetic indicators of the stem cells' location, like watching fireflies in a field at night.
Patrick Goodwill, a graduate student in the Conolly lab, has constructed an MPI scanner that they hope will develop into a device to track stem cells through animals. And because the magnetic compounds are already FDA approved, human applications may also be developed for monitoring cell therapy, drug delivery, and inflammation.

Next stop
This is by no means a complete list of the ways MRI is being tinkered with at UC Berkeley. There are also projects that involve rotating magnets and imaging of chemical reactions in real time. UC Berkeley even has a Brain Imaging Center that uses MRI to monitor brain activity. And although the majority of MRI research at UC Berkeley is non-clinical (some key collaborations do go on with UCSF and with both Stanford hospitals), Bouchard thinks that this is an unusual advantage. "Since we don't have a medical school, we tend to take a more basic research perspective, and we tend to collaborate with more material scientists and physicists," he says.

Professor Conolly says he would appreciate more contact with hospitals, but that he agrees that the basic research aspect of UC Berkeley's MRI work is advantageous in many respects. "When I was at Stanford," he comments, "the moment anyone had a new idea, they immediately wanted to start a new company. Although I miss the convenience of having the two hospitals right next door, Berkeley is doing some really pioneering research on NMR and MRI."

MRI has come a long way in a very short time, but it could go even further—with smaller magnets, or without magnets, at one tenth the current cost, or following molecules in a bloodstream instead of just following the bloodstream itself. The world's researchers are doing beautiful things with MRI, but UC Berkeley researchers are doing, well, funkier things. In addition to listening to the needs of the medical establishment, our researchers also listen to the whispers of the atoms themselves.

Natasha Keith is a graduate student in chemistry.

Comments on this article? Drop us a line at with 'letter to the editor' in the subject!