Alzheimer's Predictor

Two complementary brain scans can pick up Alzheimer's before it happens.

By Brad Kloza|Tuesday, September 19, 2006

Whether it's misplacing your keys or forgetting a phone number, lots of people wonder when simple lapses in memory might be an early sign of Alzheimer's. Now researchers are working on software that might help predict who's really at risk.

New York University brain researchers Susan De Santi and Lisa Mosconi are part of a team developing software that they say will help tell the difference between a person who's just getting old and one who's on the road to Alzheimer's. The software combines information from two different types of brain scans: MRI and PET.

The software takes advantage of the strengths of these two types of brain scans, and each helps mitigate areas where the other is weak. MRI, for instance, is a very good tool to study the inner makeup of the brain and to pinpoint precise locations. PET is much less precise, but does a good job of showing where the brain is using sugar, the energy source of living cells. Plunges in energy use could signal decline in brain function, and the researchers decided to focus on the hippocampus, a part of the brain known to be affected in Alzheimer's.

"What we are trying to do is to find a measure that would predict decline from normal aging to Alzheimer's disease," says Mosconi. "And it looks like the hippocampus is particularly involved in early Alzheimer's disease."

The researchers tracked 53 healthy volunteers for up to 14 years. Using brain scans taken at the beginning of the study, the software accurately predicted a person's future mental decline about 80 percent of the time. Specifically it predicted future Alzheimer's disease with 85 percent accuracy; future mild cognitive impairment with 75 percent accuracy; and a future of healthy cognitive function 89 percent of the time.

Albert Einstein College of Medicine neuroscientist Luciano D'Adamio, who was not involved in the project, points out that this is good, but there is still concern about the other 20 percent that the software is still getting wrong.

"So you want to, when you do a diagnosis, you want to have a diagnosis which is very efficient and precise," he says. "So you don't want to miss 10, 20 percent of the patients."

De Santi and Mosconi agree, and hope that future versions of their software will be more accurate, and might even be able to predict how long it might take for Alzheimer's to take hold.

Comment on this article