Dartmouth center collects functional MRI data

Functional MR brain imaging is gaining adherents, but the size of the studies, which commonly include a terabyte of data, poses a real challenge to informatics managers and those who want to share results with other neuroimagers.

To help facilitate fMRI data sharing, Dartmouth College has opened a clearinghouse in the form of the fMRI Data Center to collect the data produced by fMRI research.

"We're collecting the gigabytes and gigabytes of information generated during research studies in one public database so other teams can confirm results or pool information for comparison studies," said Jack Van Horn, Ph.D., the center's director.

The ability to examine primary data also allows researchers to investigate the robustness of published conclusions from different analytical methods, as well as determine statistical significance thresholds.

The repository currently houses over 50 studies, with new studies arriving at the rate of about one per week, Van Horn said. Authors whose papers have been published or accepted for publication may submit their data to the center.

"We receive studies from authors on CD, DVD, DLT, and DAT, as well as via FTP," Van Horn said.

Larger ones tend to arrive on media, smaller ones via FTP. Plans call for making studies available online in the future.

The sheer size of the data sets means that transmitting them over the Internet is impossible. Until the arrival of Internet 2 or similar megabandwidth pipes, researchers requesting studies from Dartmouth must settle for a stack of 15 or more CDs or DVDs.

"We've filled over 800 data set requests from all over the world," Van Horn said. "Researchers visit the center's Web site (http://www.fmridc.org/), surf the abstracts of the available studies, then request one that interests them, which we copy to CD or DVD and mail."

The project is funded by the National Science Foundation and other sources, and it satisfies the data sharing edict the National Institutes of Health published Feb. 26, 2003.

At first glance, the center would seem to be a welcome development. A controversy surfaced in the wake of its 2000 launch, however. Its first director, Michael Gazzaniga, Ph.D., a prominent cognitive neuroscientist, suggested that several journals - including Journal of Cognitive Neuroscience, of which he was editor - were planning to require the release of primary data as a condition of publication. This caused many in the field to squirm.

The Journal of Cognitive Neuroscience does now require fMRI data release as a condition of publication. The Journal of Neuroscience and the Proceedings of the National Academy of Sciences both strongly encourage authors to release raw data, and both are considering making it a condition of publication at some point in the future, Van Horn said.