Magnetic Resonance Imaging Captures Brain In Action

Neuroscience researchers have long wished for improved functional mapping of the brain--for the ability to create better images showing the brain in action, rather than just its structure. In the past, these researchers have depended mainly on positron emission tomography (PET), but safety concerns posed by the radioactive contrast agents involved, along with the relatively coarse resolution and slow speed of PET imaging, have limited the usefulness of their results. But now, a new, naturally occurring contrast agent that is used not with PET but with magnetic resonance imaging (MRI) is brightening prospects. Coupled with recently developed high-speed MRI techniques, this new agent is helping researchers create functional brain maps that display detail and sequential processes never before seen. The newly found contrast agent is hemoglobin, present throughout the body. MRI researchers are taking advantage of the slight difference in magnetic properties between hemoglobin carrying oxygen and hemoglobin from which the oxygen has been stripped to generate their images. Oxygenated hemoglobin is diamagnetic, and deoxygenated hemoglobin is paramagnetic, meaning that oxyhemoglobin is weakly repelled and deoxyhemoglobin weakly attracted by a magnetic field. And, although experimental MRI instruments with high magnetic fields produce the best images based on the hemoglobin contrast, many commercially available clinical MRI machines are also capable of creating such images. "A lot of what we're trying to do is to confirm and expand upon the PET results," says Ravi Menon, a physicist at the University of Minnesota Medical School's Center for Magnetic Resonance in Minneapolis. "But PET is an invasive technique--you have to inject a radioisotope--which means you cannot continuously do this type of experiment. And PET also has a much slower time resolution, typically on the order of five minutes. We have resolution on the order of seconds [with MRI], so we can follow, essentially, real-time changes." Because of the time needed to generate an image, says Menon, PET researchers often have to average images to create functional maps similar to the ones now being done with MRI, thereby losing detail. "What they have to do is, say, add the images of 10 people together in order to demonstrate motor regions being activated," he says. "But no two people are the same size, shape, or anything so the area that's activated becomes very blurry--whereas we can see, in a single person, in a single shot, what is activated." Also, the spatial resolution of MRI is about 25 times that of PET, by volume. PET can resolve an area 5 millimeters by 5 millimeters across a 5-millimeter-deep imaging slice, while MRI is able to resolve an area 1 millimeter by 1 millimeter across the same depth slice. The Minnesota group that Menon is part of has worked with researchers from AT&T Bell Laboratories in Murray Hill, N.J., conducting a number of experiments based on the new technique (S. Ogawa, et al., Proceedings of the National Academy of Sciences, 89:5951-55, 1992). Some involve visual stimulation, and some involve motor tasks or stimulation. Some even require the volunteer subjects to engage their imaginations. What the researchers are looking for is the increased presence of oxygen, as revealed by higher concentrations of oxyhemoglobin, as an indicator of increased neural activity. "We've done experiments, for example," says Menon, "where we have a subject lie in a magnet and then say to them: `Intensely visualize your house, and go through it and try to count all the wall outlets, or the light switches.' Or, `Imagine you're on your daily walk, and try to count all the stop signs.' You have to visualize in order to do this, and we get activation in various areas." Even though they are not averaging images in the same way as do PET researchers, the Minnesota team is combining the information from several MRI scans to create their images. "We do subtractions of images made during the activation from images made during rest, normal control," Menon says. "And that can then be overlaid directly on the anatomical image." The anatomical image would be a conventional MRI scan, not using hemoglobin as a contrast agent. The advances in Minnesota have their beginnings in work done by Bell Labs researcher Seiji Ogawa. Ogawa used the magnetic characteristics of oxy- and deoxyhemoglobin to study major blood vessels. One major difference between this early work and the current studies is that Ogawa was imaging relatively static structures. MRI instruments being used by the researchers mentioned in the accompanying article to create blood-oxygen-level-based images are available from the following manufacturers: * Advanced NMR Systems Inc. 46 Jonspin Rd. Wilmington, Mass. 01887 (508) 657-8876 Fax: (508) 658-3581 * Bruker Instruments Inc. [Purchased General Electric NMR Instruments] 19 Fortune Dr. Billerica, Mass. 01821 (508) 667-9580 Fax: (508) 667-3954 * Siemens Medical Systems Inc. 186 Wood Ave. South Iselin, N.J. 08830 (908) 321-3266 Fax: (908) 321-3287 Researchers at Massachusetts General Hospital, led by physicist Ken Kwong, are also using hemoglobin as a contrast agent to create functional maps of brain activity (K.K. Kwong, et al., PNAS, 89:5675-79, 1992). Kwong's group, in fact, was the first to use the technique to create movies showing the process of the brain's response to visual stimulation unfolding over time. Kwong says that Linus Pauling first discussed the paramagnetic character of deoxyhemoglobin in the 1930s. But it was only recently that the possibility of using the magnetic character of hemoglobin in MRI occurred to Kwong, during discussions with Robert Turner, a physicist at the Cardiac Energetics Laboratory of the National Institutes of Health in Bethesda, Md. Turner was conducting a series of imaging experiments in which he deprived cats of oxygen for a short period and then restored the oxygen. He found that the MRI signal from the cats' brains went down as the level of deoxyhemoglobin rose. Then, when oxygen was restored, the signal climbed again. But the surprise was that the signal went beyond its baseline level. "Instead of just going back to baseline," says Kwong, "it overshot a little bit. [The vascular system] overcompensates, bringing in more oxygen, and with more oxygen in the blood, the signal would actually go beyond the baseline." Kwong immediately saw possibilities for functional mapping of the brain's response to tasks in this rebound effect. Previous work, such as Ogawa's studies, had shown that deoxyhemoglobin dropped the signal, but here was the possibility that oxyhemoglobin would cause the signal to go up--and that the change was measurable. "I picked up that phenomenon and decided to apply it to people in activation," says Kwong. "It was very much Ken's original idea to try this out on humans," says Turner. "I was actually a volunteer in one particular experiment, visual stimulation. It was one of the most startling things I've ever seen in my life, when he showed me the movie loop of the images afterward, to see my brain like that." Turner is now doing similar experiments himself. He has an advantage over Kwong's group in that he has one of only three MRI all-body machines in the United States that uses a 4-tesla magnetic field. (The Minnesota team has one also; the third is at the University of Alabama, Birmingham, and is not used for these types of experiments.) Kwong's machine, like most clinical MRI instruments, operates at 1.5 tesla. The significance of this is that the magnetic differences between oxy- and deoxyhemoglobin are more dramatic at 4 tesla and thus more easily imaged. At 1.5 tesla, Kwong's group sees about a 2 percent or 3 percent difference between the two hemoglobin states. At 4 tesla, Turner and the Minnesota group see a 15 percent to 20 percent difference. Additionally, the relative interference from background noise is lessened, again aiding imaging. Another group, under James S. Hyde at the Medical College of Wisconsin in Milwaukee, also did important early work in developing ways to translate increased levels of blood oxygenation into measures of neural activity. Now, dozens of other groups are involved in studies of this kind. One point that both Menon and Turner emphasize is that the imaging speed is not necessarily the limiting factor with the new technique--in a sense, the human body is. Any of the several very fast MRI techniques, those able to capture images in tens of milliseconds, is able to keep up with changes in the brain's vascular system. Says Menon: "The neurophysiological changes that occur, in terms of the blood flow, take a second or two. "So, really, to go any faster than that [with imaging] is just a waste of time. You just produce a lot of data." "We can see sequential activation," using vascular response to study the brain's response to visual stimulation, says Turner. "Unfortunately, it's not quite fast enough to be able to see the kind of sequential activation that takes place as information moves from the primary visual cortex to the associated visual cortex. Some neuroscientists would love to be able to see that, but that takes place in around 100 milliseconds. The hemodynamic response, the blood flow changes in the brain, are more like five or six seconds. "It's just that the brain itself is sluggish," Turner adds. "It's not the technique by which we're observing the blood flow." Menon also says that handling and storing the huge amount of data generated by MRI imaging experiments has become a problem. "It's what limits us," he says. "We have 4 gigabytes of disk space, and it fills up very quickly." Kwong agrees and suggests that the MRI instruments themselves are capable of generating the data to construct three-dimensional images. "But that's a lot of memory," he says. "Right now, if I collect one experiment, it uses about 200 megabytes. And that's a small experiment. Even though I have an optical disk right now, I use the whole disk just for one experiment." Great Potential Overall, the new technique holds great potential in several areas of neuroscience, says Minnesota's Menon. The ability to create images of real-time processes in the brain, especially, is generating excitement. "These things have never been seen with PET," Menon says, "because the time scale was too short. In terms of cognitive studies, it's going to be important. And it's going to be useful for surgical planning." Already, researchers using the new technique have been able to generate movies in two dimensions of the brain responding to stimulation. And they have been able to generate three- dimensional static images of the brain. If demanding computational and data handling challenges can be confronted, some researchers hope that the future may hold three-dimensional movies of the brain performing such tasks as comprehension, learning, and memory.

Magnetic Resonance Imaging Captures Brain In Action

Interested in reading more?

Become a Member of