A mathematical technique used to measure shapes provides researchers with an innovative tool for studying the link between brain structure variations and several psychiatric disorders.
BY LIZ LIPTON
Since the mid 1980s, morphometrics—a mathematical way of studying shape—has been used to detect minute shape differences in MRI images of brain structures, according to Paul Thompson, Ph.D., an assistant professor of neurology at UCLA.
He noted that University of Michigan mathematician Fred Bookstein pioneered the use of morphometrics with brain images. Researchers have used this technique to study psychiatric illnesses such as schizophrenia, Alzheimer's disease, and fetal alcohol syndrome. It has also been used to study nonpsychiatric disorders involving craniofacial anomalies, differences in size and shape of spinal vertebrae, and differences in shape changes in tumors, explained Thompson.In fact, Thompson, and all the other researchers cited in this paper, have been investigators in one or more research studies using morphometrics.
"As the resolution of MRI increases, morphometrics has become increasingly valuable because smaller structures with specific functions are more readily seen," said Monte Buchsbaum, M.D., the editor of Psychiatry Research and Psychiatry Research: Neuroimaging.
Similarly, Jay Giedd, M.D., said, "Morphometrics is an important tool that enables researchers to understand neurobiological bases of disease." Giedd is chief of brain imaging for child psychiatry at the National Institute of Mental Health (NIMH).
Both Thompson and Peter Buckley, M.D., an associate professor of psychiatry at Case Western Reserve University and medical director of North Coast Behavioral Healthcare, stressed that morphometrics should not be heralded as the best method for analyzing structure but rather as one of three primary methods used to study MRI brain images, explained Thompson.
The other two techniques, he said, are "volumetrics, which derives the volume of different structures, and tissue analysis, which derives the location of gray matter, white matter, and cerebrospinal fluid or abnormalities in these tissues."
Thompson explained that "because diseases do different things to the brain, sometimes researchers want to study volume, other times shape, and sometimes changes in tissues. For instance, volumetrics is especially useful in studying Alzheimer's disease. Why? Because certain brain regions shrink, tissue is lost, and the volume decreases. On the other hand, tissue analysis is particularly well suited to studying multiple sclerosis because it can detect the demyelination of white matter."
Thompson emphasized that researchers often use two or three methods to get the most information. To illustrate this point, he noted that his Alzheimer's disease research revealed degeneration of regions of the corpus callosum through use of volumetrics, shape differences in the corpus callosum through morphometrics, and the loss of gray matter in the temporal lobe through tissue analysis.
How widely used are these techniques? "Of the three," Buchsbaum said, "volumetrics is the easiest to use and is in fact used by nearly everyone. [In contrast,] tissue analysis is more difficult to use because it requires a complex computer technique to separate the gray and white pixels. And finally, only a handful of researchers uses morphometrics—even though it's not a difficult technique." Buchsbaum is a professor of psychiatry at Mount Sinai School of Medicine.
When asked why more brain imaging researchers don't use morphometrics, Buckley replied, "Because it is relatively new and technically challenging, and because there are numerous other techniques, such as those studying function and volumetrics, that have produced compelling results."
He pointed out that "the challenges of neuroimaging research—including morphometrics—are several-fold. There is an increasing emphasis on integrating findings from studying structure as well as function. Also, there is an ongoing effort to understand symptoms and the effects of treatment. Another new development is the use of sequential imaging over the course of an illness to better understand illness trajectory."
Why Morphometrics Is Important
"Morphometrics is an important technique because it is well-suited to detect subtle shape differences that may very well not be accompanied by changes in volume and tissue—as is the case in autism," said John Csernansky, M.D., the Gregory B. Couch Professor of Psychiatry at Washington University School of Medicine.
"Several groups are working on similar techniques, and . . .techniques that are high dimensional offer the best precision and the greatest promise."
Another advantage of morphometrics, according to John DeQuardo, M.D., is that "it allows researchers to study the brain as a whole and examine spatial deformation between structures and how shape differences might affect other neuroanatomic structures, whereas volumetric techniques look at an anatomic substructure in isolation and measure the volume."
For example, said DeQuardo, the director of adult inpatient psychiatry at the University of Michigan Hospital, "in our research on schizophrenia, we examined how the deformation of the corpus callosum might affect other structures, and we found that the splenium of the corpus callosum is moved backward relative to the brain stem, which produced an apparent increased size of the quadrigeminal cistern."
When asked about future research and how findings from morphometrics research dovetails with other psychiatric research, DeQuardo explained that "in the future, researchers hope to use these anatomical shape differences to aid in the diagnosis of mental illnesses and as markers for researchers studying the genetic basis of schizophrenia." "The first step in using morphometrics," Csernansky explained, "proceeds in one of three ways: (1) highly trained researchers can trace the outline of a structure using a mouse; (2) researchers can hand-plot points on landmark locations such as those that surround the corpus callosum; and (3) researchers can use a computer program to plot thousands of points around the perimeter of the structure. The last two methods are referred to as computer-based morphometrics.
"The next step is to align these images," DeQuardo pointed out. "And because the brain shapes are different sizes in different subjects, the researchers use a statistical process that eliminates the size and orientation of the collection of points. This collection of points forms a landmark shape for each subject."
In the next step (a statistical averaging process) a computer program computes the entire population’s average shape. Then each individual image is compared with the average population's landmark shape. In doing this, researchers ultimately determine average images for all diseased and normal subject groups.
Finally, the computerized image of a grid is used as the background against which the average shape of the average patient image is mapped onto the average shape of the normal control image or vice versa. The difference in shape is shown as deformation in the grid. The procedure also produces numbers (shape coordinates) that quantify how far the landmarks have moved in this comparison.
Accuracy Questioned
Not all researchers are sold on morphometrics, however. The research group headed by Nancy Andreasen, M.D., editor of the American Journal of Psychiatry and director of the Mental Health Clinical Research Center at the University of Iowa, questioned the accuracy of morphometrics used with three-dimensional images.
"We had used the technique successfully with two-dimensional images, but when we switched to 3-D images, we were unable to find the exact same landmark point on each of the 3-D images, so we stopped using it," explained research scientist Dan O'Leary.
In contrast, Buchsbaum successfully plotted points in 3-D images in a recent study. Although he admitted there were some difficulties in reliably plotting the points on the landmarks, he believes these problems can be solved. In fact, Buchsbaum has received a grant from the National Institute of Mental Health to develop more accurate techniques for plotting these points.