The  world's  first  morphometrician  and  his  colleagues   depict  the  shape  of  the  human  brain  in  sickness  and  in  health

By Diane Swanbrow

A few years ago, Fred Bookstein was browsing in his library when an illustration in the second volume of The Life, Letters and Labours of Francis Galton by Karl Pearson (1924) stopped him cold. The picture was a geometric diagram of Galton's system for classifying human profiles. Galton (1822-1911), the British natural philosopher and polymath, was a pioneer in the racist field of eugenics, among many far more notable achievements, including the invention of weather maps and the description of fingerprints.

Galton's interest in criminology led him to devise a way to analyze profiles by measuring the distances between sets of standard facial features, such as the corner of the eyes, the tip of the chin and the bridge of the nose. Calling his method "anthropometry," Galton proclaimed it "the best means available for identifying habitual criminals."

Bookstein (pronounced Book-steen) was astonished to see in Galton's amateur methods one of the key tenets of a statistical technique he had recently developed after 10 years of work, a technique that he hoped would solve the centuries-old problem of describing how one shape differs from another.

Morphometrics is the name Bookstein coined for his technique of measuring biological shape and change, and he considers Galton to be one of its godfathers. But the field has come a long way from Galton's pernicious, early-20th century assumption that a "normal" non-criminal face belongs to a white northern European male, probably an Englishman. "Like almost every educated Brit, Galton was a racialist," Bookstein says. "His variety of eugenics was more genteel than the virulent form that erupted later in the century. But the irony is that eugenics didn't fall because of politics but because it was just plain bad genetics."

Still, it has taken most of this century to heal the scar of eugenics, according to Bookstein. Today the most promiosing application of morphometrics is not to human faces at all, but to human brains. And Bookstein and others are extremely careful about just what they mean by terms like "average" or "normal."

Before morphometrics, clinicians learned about variations in the shape of the "normal" brain in the course of their experience. But this knowledge could not be represented visually. Each anatomy text showed only a single version of the "normal" brain, based on the sliced brain of a single cadaver or the magnetic resonance image of the brain of a single patient. The "normal" brain, however, comes in a range of shapes and features.

"The range of normal could not be depicted because we had no ways of squeezing objectively defined 'average pictures' out of stacks of individual images, each of which a clinician was willing to call 'normal,'" says Bookstein, a Distinguished Research Scientist at the University's Center for Human Growth and Development and the Institute of Gerontology. Bookstein figured out how to construct a "normal" image as an average derived from billions of high-speed computer computations on images of many different individual brain images. Now, with morphometrics, digitized images of individual brains can be speedily and precisely compared with a computerized image of an average brain. The composite image will reflect the current, agreed-upon range of "normal" brain structures and shapes.

Bookstein's technique has already yielded some intriguing clinical findings. This past summer John DeQuardo, a U-M psychiatrist, used software developed by Bookstein and mathematician William D.K. Green of the Center for Human Growth and Development, to identify previously unsuspected structural abnormalities in the brains of schizophrenics.

While the links between various anatomic features and the symptoms of disease remain unclear, previous studies have suggested that the brains of schizophrenics are a bit small, with enlarged cerebral ventricles (the brain cavities that connect with the central canal of the spinal cord). But those studies were limited in scope, investigating one or another region of interest. As a result, they begged the hotly debated question of whether the extent and focus of detected abnormalities are local or global.

DeQuardo deployed the new technique of morphometric analysis to show that in schizophrenics, the abnormalities are both global and local. "The brains of schizophrenics are significantly smaller than the control average" DeQuardo says, "and the corpus callosum the hard body connecting the two hemispheres of the cerebrum [the gray matter that allows people to think and speak], appears thinner than normal all along its length."

According to Rajiv Tandon, associate professor of psychiatry and director of the U-M schizophrenia program," John's application of Fred's image-averaging technique nicely complements traditional region-of-interest analyses, which are nearing a dead-end. It's very useful to be able to step back and view the brain whole, since what's wrong in schizophrenia is probably not just one abnormality, but many."

Down the road, Tandon speculates, morphometric analysis may allow researchers to sort out which brain abnormalities are developmental and which are degenerative. These differences, he notes, may be related to various behavioral dimensions of the disease, from delusions and hallucinations to a loss of interest in life's normal pleasures.

DeQuardo presented the findings based on morphometric analysis of 14 undiagnosed controls at a national psychiatric conference last May, and an article on his findings is being reviewed by Psychiatry Research: Neuroimaging. Now he and Bookstein are working to replicate the analysis, comparing a news group of 25 schizophrenics to a new group of 25 controls.

Bookstein plans to present the latest results, whatever they are, at a symposium on research progress in the Human Brain Project, which was launched last year to give scientists the tools they need to access the vast amounts of information on the brain now accumulating. The $6 million project, which is supporting the research of Bookstein, Green and more than 40 other investigators from around the world, is sponsored by the National Institute of Mental Health (NIMH), the National Institute on Aging, the National Institute on Drug Abuse and other federal agencies.

"I hope we'll replicate the schizophrenia findings," Bookstein says. "If we do, the value of morphometrics as a tool of scientific analysis, a means of achieving new insights into the brain, will be confirmed. If we don't, well, I'll be disappointed but not defeated. Whenever you do something new, you take the risk of being wrong. You can't expect a tool, even one as mundane as a screwdriver or a voltmeter, to work every time you use it "

Whether or not morphometrics works this time out, Bookstein's accomplishment is substantial. The 1994 edition of the "bible" of mathematical statistics, The Advanced Theory of Statistics, refers for the first time to "Bookstein's (1986) shape variables." And DeQuardo and other clinicians are using morphometrics to help evaluate their patients' brains.

Morphometrics is also expected to play a crucial role in solving the puzzle of the human brain, according to Stephen Koslow, director of the division of neuroscience and behavioral science at NIMH and coordinator of the consortium of federal research organizations that are funding the Human Brain Project.

"Other studies are providing pieces of the puzzle," Koslow says. "They're looking at one set of chemicals, one electrical system, one piece of the brain's structure or another. Yet the brain works as a unified whole. Fred's achievement is that he's found a way, by warping brain images, to create an aggregate brain and then to compare one brain with that aggregate. He's found a way to see the brain whole."