Looking Deep into the Body

On October 3, 1971, engineer Godfrey Hounsfield of the U.K. electronics company EMI and neuroradiologist James Ambrose waited nervously at Atkinson Morley Hospital in London. Two days before, they had performed the first computed tomography (CT) scan on a 41-year-old woman with a suspected brain tumor. Hounsfield and Ambrose had then dispatched the data from the scan to EMI’s facility on the other side of London so that the company’s mainframe could analyze it. The pair was itching to see the results.

Hounsfield had first proposed a device that could see into the brain three years earlier. On a traditional x-ray, the organ appears as what one scientist called “a gray, undifferentiated fog.” But Hounsfield argued that using x-rays to image individual layers, or slices, of the brain would clear the murk and resolve previously hidden details. Like a person reading a book, his system “would be capable of extracting the information from one page (or slice) only.” The patient had been scanned by a prototype device based on Hounsfield’s ideas.

When the EMI computer’s image of the woman’s brain finally arrived at the hospital, it made both men “jump up and down like football players who had just scored a winning goal,” Ambrose recalled. “There was a beautiful picture of a circular cyst right in the middle of the frontal lobe,” Hounsfield wrote. Surgery revealed a growth known as a cystic tumor in the location indicated by the image.

For devising the first workable CT scanner, Hounsfield earned the 1975 Albert Lasker Clinical Medical Research Award. He split the prize with neurologist William Oldendorf of the University of California, Los Angeles, who had crafted a rudimentary brain-scanning device a few years earlier. Hounsfield also shared the 1979 Nobel Prize in physiology or medicine.

Hounsfield’s accomplishment was doubly impressive because “he had to envision that this would work in human heads and then build it to show that it worked,” says medical physicist Cynthia McCollough of the Mayo Clinic. His work also stands out, says Mayo Clinic radiologist Richard Ehman, “for how quickly these advances were translated to clinical practice”: a mere three years. By contrast, a new drug now takes an average of 15 years to reach patients, he says.

Lasker winners also catalyzed the development of two other transformative imaging techniques: positron emission tomography (PET) and magnetic resonance imaging (MRI). Biochemist Louis Sokoloff of the National Institute of Mental Health, who won the Albert Lasker Clinical Medical Research Award in 1981, didn’t invent the PET scanner. But “his work was instrumental in translating PET from a research tool into a clinical imaging modality,” says David Townsend, a physicist and founding director of the Clinical Imaging Research Centre at the National University of Singapore.

Research by the winner of the 1984 Albert Lasker Clinical Medical Research Award, biochemist Paul Lauterbur, laid the groundwork for MRI scanners. Lauterbur, then at the State University of New York at Stony Brook, made “the almost inconceivable leap . . . that you could create images using magnets,” says Daniel Sodickson, chief medical scientist at the diagnostic testing company Function Health. Lauterbur also shared the 2003 Nobel Prize in physiology or medicine.

CT, MRI, and PET are now performed hundreds of millions of times each year in the United States. “It’s hard to overstate the impact these techniques have made,” says radiologist William Dillon of the University of California, San Francisco. They haven’t just changed medicine, but by revealing so much internal detail they have changed how we view ourselves, giving us “a more intimate, direct relationship with our bodies,” Sodickson says.

Slicing the Brain—Virtually
When Oldendorf began his research in the late 1950s, available methods for imaging the brain were often grueling for patients and uninformative. X-rays of the head weren’t much help for diagnosing problems such as brain tumors. An invasive alternative known as pneumoencephalography involved injecting air into the lower spine. In one version of the procedure, patients were strapped into an apparatus “sort of like the Tilt-A-Whirl” amusement park ride, Dillon says. By flipping patients upside down and tipping them back and forth, the device maneuvered the injected air so that some structural details of the brain showed up on an x-ray. The experience was unpleasant—and sometimes fatal.

A 1958 conversation with an engineer from a food company inspired Oldendorf to develop a less traumatic method. The engineer was trying to devise an automated x-ray system to identify oranges with internal frost damage. Oldendorf realized that both were struggling with the same problem: how to see inside an object. He wondered whether “perhaps by a more elaborate scanning pattern and by more sophisticated analysis of the absorption of an x-ray beam, the actual structural pattern of the interior of the head could be defined.”

Another researcher described Oldendorf as “a good all-around tinkerer,” and within two years he had cobbled together a DIY prototype. From a model train set he borrowed a section of track and a flat car. The platform for the specimen was the turntable of an old phonograph. He built the device’s radiation shield from lead he had melted on his kitchen stove.

Oldendorf’s homemade apparatus fired a narrow beam of gamma rays toward a detector. As the rays pass through a specimen, they weaken as they are absorbed or scattered. By analyzing measurements taken from different vantage points, Oldendorf thought he could reconstruct the specimen’s internal structure. In his prototype, the detector stayed in the same position while the specimen moved, sliding slowly along the train track and simultaneously rotating.

To test the apparatus, Oldendorf trained it on an ersatz head that consisted of a block of plastic sporting two concentric rings of iron nails. That double layer represented the skull, whereas one iron nail and one aluminum nail in the center of the rings simulated brain features. Oldendorf reported in a 1961 paper that the prototype could detect the aluminum nail through the mock skull.

With that success, he patented his device and approached companies that made x-ray machines about developing it further. They turned him down. “The idea was rejected because the manufacturers could not imagine a viable market for such a device,” he wrote. In fairness to those companies, his invention was a long way from a usable scanner. For one thing, he hadn’t shown that it could produce an image. “Oldendorf recognized the problem” of visualizing the brain, says radiation physicist Norbert Pelc of Stanford University, but he didn’t crack it.

Hounsfield did. Although he was unaware of Oldendorf’s work, their approaches were similar in some ways. Hounsfield jury-rigged his first prototype from a lathe and other spare parts. He used it to scan inanimate objects and tissue samples. A source of gamma rays gradually tracked laterally along the specimen as detectors measured the strength of rays that penetrating the material. After each pass, the specimen rotated slightly, and scanning began again.

The procedure’s next step set Hounsfield’s approach apart. He devised an algorithm to analyze the data captured by the detectors and generate an image that showed a slice of the brain. “He solved it computationally using the measurements he made,” Pelc says.

Still, the device’s first version was impractical, requiring nine days to complete a scan. Switching to x-rays, which can produce a more intense and controllable beam, shrank the time to nine hours. After testing the modified device on 15 brains removed from cadavers, Hounsfield and colleagues decided to build a machine that could image a living person’s brain in a few minutes. The device was ready for the first patient just over 18 months later. The scanners caught on quickly. “As soon as clinicians saw the images, it was all over,” Pelc says.

This Is Your Brain on Glucose
Sokoloff might never have started the research that led to diagnostic PET scanning if he hadn’t taken a yearlong sabbatical to Paris in 1968. When he returned to the United States, the project on thyroid hormones that his lab had been pursuing for several years “was in chaos,” he recalled. He decided to study a different question that had long intrigued him: how metabolism varies from place to place in the brain.

Hungry brain cells slurp up glucose and metabolize it to generate energy. Scientists thought that if they could determine how much glucose different parts of the brain used in different situations, they could better understand the organ’s regional specialization and probe how diseases affect its function. “People knew that glucose was the fuel for the brain, but they didn’t have a way to measure it regionally,” says Gerald Dienel, an emeritus biochemist at the University of Arkansas for Medical Sciences who worked in Sokoloff’s lab between 1984 and 1996.

Sokoloff and colleagues developed a technique that relied on 2-deoxyglucose (DG), a type of glucose that cells absorb but can’t break down. Sokoloff first heard about DG in the late 1950s and “filed the idea away,” he wrote. When he finally started investigating the molecule, he wondered whether scientists could use it to track metabolism in the brain. It was “a crazy idea,” he admitted, because DG inhibits metabolism. But he and his team suspected that DG got stuck in tissues that absorb it, so it could serve as an indicator of metabolic rate.

After injecting animals with radioactively labeled DG, the researchers found that the molecule revealed how different parts of the brain used glucose. Their results, presented at conferences in the mid-1970s, wowed other scientists. The method “was a major advance for animal studies,” Dienel says.

Then one of Sokoloff’s colleagues asked the obvious question: Could they modify the technique to image the human brain? “I didn’t see how,” Sokoloff recalled. The version of DG they enlisted was tagged with radioactive carbon-14, and the technique for visualizing it in the brain wouldn’t work in a living animal, let alone a person. However, some of Sokoloff’s collaborators found that affixing radioactive fluorine to DG, yielding fluorodeoxyglucose (FDG), produced a tracer suitable for humans. Those scientists first tried to detect FDG in a person’s brain with a device known as a single-photon tomographic scanner, but they quickly discovered that PET scanners “offered better spatial resolution and accuracy,” Sokoloff wrote.

Scientists had identified other tracers for visualizing brain metabolism, but they were useful mainly for academic research, says medical physicist Terry Jones of the University of California, Davis. Sokoloff’s work allowed the use of FDG to image tumors, which actively metabolize glucose. FDG’s value for detecting and staging human cancers “made commercialization of PET technology possible,” Jones says.

Seeing with Magnets
Lauterbur’s key insight came one evening in 1971. He had just watched a researcher show that the chemical analysis technique called nuclear magnetic resonance (NMR) could distinguish between normal and cancerous tissue. However, the tissue samples came from dissected rats, and the experiment put Lauterbur off. A similar noninvasive approach would work, he speculated, if measurements could be taken “from outside the living body with sufficient spatial resolution.”

That catch was that NMR doesn’t reveal spatial information. The technique works by exposing a sample to a strong magnetic field, which causes the spins of protons in the sample to line up. Prodding those protons with radiofrequency (RF) radiation spurs some of them to “become little radio transmitters,” Sodickson says, and they release RF radiation at specific frequencies. By analyzing those signals, researchers can infer the sample’s composition. But they can’t tell where in the sample particular protons are.

NMR applies a uniform magnetic field to a sample. Lauterbur’s epiphany was that a magnetic field that differs in strength from place to place could elicit the spatial information needed for imaging. The frequencies that a proton emits depend on the strength of the magnetic field it is exposed to. Therefore, he reasoned, signals produced by protons in a varying magnetic field could reveal their locations. “That took a lot of cleverness. It was a brand-new idea,” Sodickson says.

But Lauterbur still had plenty of work to do. He first devised an algorithm that could transform measured RF data into an image. He then calculated that it was possible to build a magnet large enough for the job. Lauterbur also performed some experiments showing that an NMR machine could image glass tubes filled with water. “It then appeared that all the requirements could be met if the right research and development could be done, so that a new and useful medical diagnostic tool would be available,” he wrote. The clunky name that Lauterbur coined for the approach—zeugmatography—never caught on. But MRI did. The first scanners reached hospitals in the early 1980s.

Because most protons in our tissues reside in water molecules, “what ends up coming out of MRI is a water map of the body,” Sodickson says. The technique “has fundamentally changed the way we practice medicine,” says medical physicist and radiologist Bruce Rosen of Massachusetts General Hospital. Lauterbur’s accomplishment, he says, was to “take what had been a test-tube chemist’s tool” and turn it into a new imaging method.

CT, PET, and MRI made a much bigger impact on medicine than Hounsfield, Sokoloff, and Lauterbur ever imagined. The techniques have increased the speed and accuracy of diagnosis, improved doctors’ ability to deliver and monitor treatments, and spared patients from many risky procedures. MRI, for example, has made a big difference for people with multiple sclerosis, in which the immune system attacks the brain and spinal cord, causing lesions known as plaques. Before MRI, confirming that someone had the disease was difficult, says radiologist Jeffrey Weinreb, former director of the MRI Service at Yale New Haven Hospital. But once the scanners were available, “suddenly, you could see plaques or defects in the brain and be more definitive about the diagnosis.” Meanwhile, CT scans, which are fast and easy to perform, have become “indispensable for the emergency department,” McCollough says.

All three techniques have become much faster, more accurate, and more powerful in the last few decades, and researchers have extended their capabilities. For example, a version of MRI known as functional MRI that can image dynamic brain activity “has revolutionized our understanding of how the brain works,” Ehman says. Although PET and CT technologies were originally developed for imaging the brain, patients now climb into machines that can scan the whole body. Because each type of technology boasts advantages and disadvantages, Pelc says, they complement one another. “I’m just happy we have all of them.”

By Mitchell Leslie