3 July 2003. The 9th annual meeting of the Organization for Human Brain Mapping was held from June 18-22 in New York City. Over 2,000 attendants escaped the rain to absorb workshops, symposia, and posters covering everything from the physics of image acquisition and imaging receptor binding to patterns of activation in the aging brain. This year’s keynote Talairach lecturer was Eric Kandel of Columbia University, New York, who talked about the molecular biology of learning, long-term memory formation, and recovery strategies. This report summarizes advances in basic imaging technologies and applications to Alzheimer’s not recently reported on this website.
Recent Developments in Imaging Techniques
In 1992, Seiji Ogawa, then at Bell Labs, demonstrated that MRI could be used to detect tiny differences in the magnetic properties of oxygenated and deoxygenated blood. Researchers were quick to exploit this technique as a way to study brain activity, and soon an explosion of studies followed examining the blood oxygenation level dependent (BOLD) response in just about every task imaginable.
More than a decade later, measuring BOLD activation still seems to be the workhorse of choice for much of the field, despite its low temporal resolution and the inherent ambiguity of inferring neural activity from a signal that reflects a combination of blood flow, oxygenation, and volume. The buzz at the meeting, however, could be felt in sessions that presented new developments in imaging techniques that may soon provide alternatives to conventional BOLD. Below are some highlights:
Current sense imaging:
Researchers from Peter Bandettini’s group at NIMH showed results of an ultrasensitive MR technique that is capable of detecting the electrical current given off by active neurons. Natalia Petridou showed that when they placed a dish of cultured rat neurons in the magnet, they were able to detect transient changes in magnetic field, presumably induced by the spontaneous electrical activity of the neurons. Although still in its infancy, this is an exciting step towards development of a noninvasive way to measure brain activity directly.
Near infrared spectroscopy:
Martin Wolf of the University Hospital, Zurich, presented data collected from visual and motor cortex using the relatively new technique of near infrared spectroscopy (NIRS). This technique involves shining infrared light through the skull and measuring the resulting reflected pattern at a sensor positioned a few millimeters away. Wolf demonstrated that the system provides valuable markers of brain activity. It is capable of detecting fast changes in light scatter, presumably due to changes in membrane properties that occur when neurons fire, as well as slower hemodynamic changes. Thus, it provides information not only about neuron activity, but also larger-scale oxygen consumption and blood recruitment. Another beauty of this technique is that it consists of little more than a lightweight headband attached to a computer by a cable. Subjects are free to move around, and imaging can be done anywhere, even at the bedside. In his keynote lecture for the session, Bruce Rosen of Massachusetts General Hospital pointed out that this technique would be useful for imaging subjects who can’t spend long periods confined immobile inside a noisy magnet or PET machine. Rosen mentioned young babies as an example, but this could also prove useful for severely ill or demented Alzheimer’s patients.
High-resolution MRI neuroanatomy:
Demonstrating the growing potential of using MRI for neuroanatomy, Rosen also showed a spectacular high-resolution image of a section of ex-vivo rat cortex imaged using an extremely high field strength (7 Tesla) magnet. You don’t absolutely have to have such an exquisite instrument, however. Simon Eickhoff of the Research Center of Jülich, Germany, gave a presentation showing that good anatomical data could be obtained even with a standard 1.5 Tesla clinical magnet. This group took images of the same human brain slices first in vivo using T1 weighting, then postmortem with T2. They then stained sections of the tissue for myelin and for cell bodies, and digitally compared the images gathered with the four different methods. For each method, they stepped along the edge of the brain, quantifying the gradations of lightness and darkness of the brain tissue at each step. Then they compared these quantitative profiles with the profiles obtained by the other methods to find the best match (defined as the pair of methods with the minimum Euclidean distance between the two sets of profiles). It turned out that standard T1-weighted in-vivo images corresponded well to myelin-stained images, and even better to composites of myelin and cell stains. Eickhoff concluded that in-vivo MR images yield anatomical data that is in good agreement with those obtained by conventional postmortem techniques. Similar structural MR images have been used to monitor the state of gray and white matter in several longitudinal studies of neurodegenerative progression in AD (see ARF related news story).
Diffusion tensor imaging is another development in white matter imaging that is gaining momentum. This technique exploits the fact that water molecules in the brain do not diffuse uniformly, but are confined by the cell membranes of neurons and therefore diffuse preferentially down axons. While this imaging technique can pick out white matter tracts, it was largely used for qualitative description or, in the words of Derek K. Jones of the National Institute of Health, “for making pretty pictures.” Jones showed that by quantifying the diffusivity and anisotropy of the water molecules at each voxel, algorithms could be used to trace the path of least resistance, making the drawing of tracts a quantitative, objective procedure. Using this procedure, Jones and colleagues traced different nerve tracts to the frontal cortex in both schizophrenics and normal controls, and found significant differences in both diffusivity and anisotropy between the two groups.
Stephane Lehericy and colleagues from the University of Minnesota showed that similar path-finding algorithms were able to trace the connections between different compartments of human basal ganglia to different cortical areas. The resulting connection maps bore a striking resemblance to those obtained by traditional neuroanatomical labeling in monkey brain. Although previous studies have used diffusion tensor imaging to show general global changes in diffusivity and anisotropy measures in AD, such new analysis tools may allow the investigation of specific connections at a level of detail previously not possible.
Functional Imaging of Alzheimer’s Disease
One major challenge in AD imaging lies in finding a way to use functional imaging as an early detection tool. Researchers are searching for definitive cognitive tests that activate those brain areas most affected in AD, such as the prefrontal and medial temporal areas. The hope is that a test that detects and monitors functional alterations in those brain areas can eventually be used to predict disease onset.
Georg Grön of the University of Ulm, Germany, showed that in a memory task that tested subjects repeatedly on the same sequences, early AD patients learned more slowly than people with depression, and both groups performed more poorly than normal controls. When these groups were imaged, the AD patients showed lower activation in hippocampus and frontal cortex than did either of the other groups.
Serge Rombouts of Vrije University Medical Center, Amsterdam, also showed that patients with mild AD have lower levels of activation than do normal elderly subjects in left medial temporal lobe structures during a picture-naming task. This suggests that the pattern of brain activation during such tasks can distinguish between normal aging and AD. He also presented evidence that patterns of functional activation could be used to detect the differences among different variants of dementia. They imaged AD and frontotemporal dementia patients with similar symptoms while performing the 2-back test, a classic behavioral probe for working memory. They found that AD patients showed higher activation in prefrontal cortex than did dementia patients, although, curiously, the FTD patients had higher activation in cerebellum.
Several presentations showed higher rather than lower activation in early AD. Susan Bookheimer of the University of California, Los Angeles, presented maps of hippocampal activation during a memory task conducted on ApoE4 and E3 carriers. She showed that the E4 carriers, while performing just as well on the task, actually showed higher activation in entorhinal cortex during learning, and that this higher activity was a predictor of whether or not subjects would eventually convert.
Randy Buckner of Washington University in St. Louis stressed that increases in general activation were commonly observed in early Alzheimer’s, and that cortical function is well-preserved in many areas. He proposed that one reason why cognitive function is not more impaired in patients with frontal decline is that other areas are recruited to compensate for any degeneration, thus leading to the seemingly illogical increases. As evidence, he cited studies showing increased activity in homologous areas following strokes or aging, and presented preliminary data from his own laboratory showing a positive correlation between amount of new cortex recruited and performance on logical memory. During the discussion session, questions arose as to whether increases in activation could really be attributed to new cortex taking over function, or whether they simply reflected the disease (see ARF related news story).
The jury is still out on whether higher or lower activation accompanies the early stages of Alzheimer’s, but everyone agreed that in order to be effective as a predictive tool, any test must be able to detect differences in activation within an individual, and not rely on statistical power from group testing. One hindrance to these efforts is the lack of tasks that yield robust activation in areas most affected by AD. For example, the hippocampus is a notoriously problematic area to activate, so developing a hippocampal-selective task may be the first step to finding a reliable imaging marker. While no consensus emerged from this meeting about a definitive marker, advances in technology and analytical methods hold promise that new approaches may not be far off.-I-han Chou.
I-han Chou is a science writer based in Japan.