Magnetic-Resonance Imaging Linked With Supercomputing Produces a 3-D Picture of Brain Activity Within Minutes.
PITTSBURGH -- Scientists at Pittsburgh Supercomputing Center, Carnegie Mellon University and the University of Pittsburgh Medical Center have teamed up to create a new capability for viewing the brain during mental activity. By linking an MRI scanner with the CRAY T3E supercomputing system, scan data from a subject's brain can be processed faster than the scanner scans, making it possible to see a realistic 3-D image of the brain while the subject is in the scanner. The researchers demonstrated this capability yesterday at Supercomputing '96, the annual supercomputing conference, held this year in Pittsburgh.
"With the T3E, we have enough computing power to stay ahead of the scanner," said PSC neural scientist Nigel Goddard. "The same processing that used to take more than a day on workstations we can now do in minutes."
"This is a big step forward," said Dr. Jonathan Cohen, codirector of the Laboratory for Clinical Cognitive Neuroscience, a joint venture of the University of Pittsburgh and Carnegie Mellon University. "The CRAY T3E eliminates the data bottleneck in this research. Ultimately, this on-line capability will make it possible to use brain-mapping as a clinical tool in diagnosis and treatment of brain pathology."
Using a technique known as functional MRI (fMRI), Cohen and his colleagues do "brain-mapping" experiments that produce high- resolution images showing what brain sites are active during different kinds of mental activity. fMRI generates a great deal of information quickly (half a gigabyte or more per experiment). In the past, computing couldn't keep up with the data, and it took 24 hours or more to produce a usable image.
To deal with this bottleneck, the researchers turned to PSC. CMU statistician William Eddy and UPMC physicist Doug Noll worked with Goddard and PSC research programmer Greg Hood to integrate and "parallelize" the software operations. This allowed them to effectively use the CRAY T3E, a "highly parallel" computing system in which hundreds of processors team up to divide the computing.
During yesterday's demonstration, a volunteer subject in an MRI scanner at the University of Pittsburgh Medical Center observed a whirling visual pattern. After about three minutes, the scanner sent data over a high-speed network to 128 processors of the CRAY T3E at PSC's machine room in Monroeville, Pa., which sent it back in about two minutes as a 3-D realistic picture of the subject's brain -- visualized on the convention floor with an SGI Onyx Reality Engine -- showing what areas "lit up" while the subject looked at the whirling pattern.
The next step, said the researchers, is to send data from the scanner to the T3E in shorter bursts, of from six to ten seconds, making it possible to produce a 3-D image almost instantaneously. This capability, said Goddard, eventually will lead to using fMRI brain-mapping clinically, such as in planning for precision neurosurgery to treat cognitive dysfunctions mapped to a specific brain site.
More information (and graphics) about this research is available on World Wide Web:
The Pittsburgh Supercomputing Center, a joint effort of Carnegie Mellon University and the University of Pittsburgh together with Westinghouse Electric Corp., was established in 1986 by a grant from the National Science Foundation, with support from the Commonwealth of Pennsylvania.