"What we want to do is to build an understanding of the network of correlations and transmissions between the various elements of the brain as it goes about performing tasks," said Stephen Hanson, chairman of Rutgers-Newark's psychology department and director of the university's RUMBA (Rutgers University Mind/Brain Activity) project.
The Hewlett-Packard Itanium II Workstation was provided as a grant – which required that both stringent research and educational criteria be met – to the RUMBA project. It will be linked with the Rutgers psychology department's newly acquired, highly powerful 28-node Opteron computer cluster and a fiber-channel data storage system capable of holding a trillion bytes of information. As a result, analyses of the immense amounts of data generated by human brain scans that would previously have taken months to complete can now be finished in days, according to Hanson.
The grant will dramatically enhance scientists' ability to analyze what is going on the brain and why.
For example, Hanson said, a series of simultaneous vertical and horizontal scans of selected cross-sections of an active human brain produces images of small cubic areas of tissue that can then be studied in detail. However, monitoring a single second of brain activity in this way typically results in 300,000 to 500,000 variables. Before investigators can begin to postulate why the brain handles a task in a particular manner, they first need to collate all of these variables to get a clear picture of how they are interacting and affecting the mind and body's responses.
Without the kind of supercomputing capabilities that Rutgers-Newark – which will share RUMBA's new ability to crunch gigantic amounts of data with the University of Medicine and Dentistry of New Jersey (UMDNJ) – now possesses, this is a daunting if not impossible task. According to Hanson, Rutgers-Newark is now only one of a handful of universities on the East Coast capable of digesting this quantity of information at anything approaching this speed. And, he adds, every byte is needed.
"If we were to scan someone's brain each second during the 10 seconds it took that person to walk over to a chair and sit down, we would end up with millions of variables," Hanson said. To monitor a human brain over the period of an hour, second to second, would take vast quantities of computational power.
The grant means that Rutgers-Newark researchers can spend less time involved in tedious number-crunching and more time looking at the data to figure out why the brain is using certain areas in conjunction with others to accomplish things.
The new combined supercomputing cluster also means that scientists can better isolate and look more closely at portions of the brain called Brodmann's areas, which are devoted to specific tasks, such as motor skills.
Analyzing the brain is made still more complex by the fact that, like fingerprints, no two human brains are alike, and even a millimeter's difference in the positioning of a particular brain structure from one person to the next can make a huge difference when trying to create a generalized brain atlas – a task that Hanson believes is futile.
And depending on how a person uses his or her brain, the Brodmann's areas can also change over time. "The motor skills region of someone like [retired tennis champion] Pete Sampras that are related to wrist and elbow reflexes and movement could be larger than those of other people because of the demands that he has put on them," Hanson said. "To say something about an 'average' Brodmann's area is very misleading."