Your brain is a complex structure, a vast network of neurons responsible for thought, feeling, impulse. Michael Greicius, assistant professor of neurology, has long been fascinated with the mysterious workings of the brain, calling it “the organ that talks back to you.” Now Greicius and his co-workers have discovered differences in the brain networks between people with Alzheimer’s disease (AD) and healthy controls—differences that may soon lead to easier diagnosis of the disease.

Networks of all kinds work best when they include many hubs, such that data, people or other elements can zip between them. This networking structure is called “small-world” and occurs in many areas of life, including our own brains.

The hubs in small-world networks aren’t necessarily close to one another, but they can be reached from other hubs through just a few steps, making flow more efficient. Take, for example, the path of news from a small town in the Bay Area, such as Vallejo. A story from Vallejo might be reported by the media hub in San Francisco, and perhaps picked up and reprinted by media hubs in New York or internationally. The news doesn’t travel from that Bay Area town to New York through every small town media outlet in between.

Greicius compares this news flow to the movement of signals through the brain. The movement of these signals forms a map of the brain network, a network that operates in small-world fashion. There are hubs through which these signals travel, including a handful of critical hubs that connect otherwise isolated signal networks.

Greicius and his colleagues hypothesized that in people with AD, the organization of the brain network would be abnormal. They studied the brains of 21 patients with mild AD and 18 controls using functional magnetic resonance imaging (fMRI), which measures brain oxygen levels to track neural activity.

The researchers looked at two indicators of interest: the richness of local connections to hubs in different regions, and the length between hubs. In subjects with mild AD, the length between hubs was not any longer than in that of the non-Alzheimer’s controls, meaning the brain signals were not slowed down by having to travel further distances. However, the hubs in some regions were not as richly connected to local networks as those in controls. Greicius likens the situation to a news flow in which San Francisco receives degraded information from Vallejo and passes that degraded information on to New York and the world. The message is still passed from hub to hub in the same number of steps, but the message is warped.

Unlike in many previous studies, the scientists looked at the whole brain and not just certain regions. The comprehensive view led them to observe that although AD patients had lower local connectivity in the hippocampus region of the brain, their prefrontal cortex region actually displayed great connectivity. Greicius believes this increase might be the brain’s way of compensating for deficits elsewhere.

The study identified 72 percent of AD patients correctly and 78 percent of controls correctly. While Greicius calls these results “encouraging,” his team is working to bump up their accuracy to the 80-85 percent range, enough for the test to be useful as a diagnostic tool.

Models created from fMRI data promote richer understanding of the brain, says biomedical informatics graduate student Kaustubh Supekar. “They will help with predicting outcomes to treatments.”

Supekar was first author of the paper reporting these findings; associate professor of psychiatry Vinod Menon, assistant professor of radiology Daniel Rubin and professor of biomedical informatics Mark Musen also contributed. Their results were published in the June 2008 issue of Public Library of Science—Computational Biology.

The scientists are now repeating the fMRI experiments on subjects with more advanced AD cases. Greicius expects to find fewer functional hubs and longer distances between hubs in this group, a sign that advancing disease further disables the brain network.