Spring 2003
|
The woman is modeling a dense-array EEG designed by Tucker and others at EGI. It contains twice as many contacts as an earlier model, allowing for more accurate measurement of brain activity, and thus a better understanding of what the brain is actually doing.
This and other work being done in this new frontier has practical applications in both research and the medical field, in particular in treating strokes, presurgery planning, neonatal monitoring, epilepsy treatment, and other neurological procedures.
Electrical Geodesics Inc., which recently received FDA approval to sell its neuroimaging equipment and software for medical use, is a direct spinoff from research undertaken at the UO psychology department's Brain Electrophysiology Laboratory. It also has close ties to the university's Brain, Biology, Machine Initiative, bringing together the UO's top scientists in the fields of cognitive neuroscience, molecular biology, genomics, and computational science.
The company licenses technology from the UO, designing, producing, and selling electrophysical neuroimaging equipment and related software. Employing twenty-eight undergraduates, graduate students, and postdoctoral fellows, it also serves as a training ground in the field of cognitive neuroscience.
"In 2002," EGI Director Don Tucker says, "sales reached $4.5 million, an increase over 2001's $4 million. We see plenty of opportunity for more growth."
You can pretty much bet that wherever any type of advanced scientific research is involved, so is a computer. EGI is no exception. It relies on advanced computer work courtesy of the Neuroinformatics Center (NIC), also located in the UO's Riverfront Research Park.
"It's fortunate the University of Oregon has its experts in parallel computing, as it is essential to what we are doing here," says Tucker.
Parallel computing, according to NIC Director Allen Malony, is the use of multiple computer processing units (CPUs) in the execution of an application program for purposes of improving its performance. By executing the calculations involved in the application "in parallel"—i.e., simultaneously on the multiple processors--the total time to execute the program can be substantially reduced. If good parallelism is achieved, it is possible to increase the number of CPUs to further achieve performance improvements.
NIC is working now to create a high-resolution model of human head tissues, which will be used in the analysis of brain function by integrating information from EEG and MRI sources. This will enable researchers and doctors to see a far greater range of brain function taking place over a longer period of time—which is one of the computational challenges.
"This is a computational problem that is difficult to solve," says Malony.
When neuroimaging measurement data are merged from two sources, Malony explains, not only is the amount of the data quite large for average experiments, the complexity of data associations increases.
"You're suddenly dealing with large amounts of data," he says.
The goal is to be able to connect the phenomena being observed with the sources in the brain of that activity, and then observe it over time.
"We want to be able to develop a computer environment able to do 3-D brain modeling in a manner capable of producing high resolution spacial representation of the brain," Malony says.
"What we're doing here is on the border between basic research and business," Tucker says. "It's exciting."
| Inquiry © 2003; University of Oregon Office of Research, Eugene OR 97403
Questions? Commments? Contact Jim McChesney, (541) 346-3017 |
 |