Sharing functional magnetic resonance imaging (fMRI) data is quickly becoming a reality

May 11 2004

The dream of saving and sharing functional magnetic resonance imaging (fMRI) data is quickly becoming a reality, according to Dartmouth researchers who run the fMRI Data Center, which archives and distributes the raw data from studies that track brain activity using fMRI. The Dartmouth researchers wrote an essay in the May issue of Nature Neuroscience about the initial reluctance and gradual acceptance of the center, and they describe the many attributes a center such as theirs offers the scientific community.

"The fMRI Data Center was created with sharing information in mind," says John Van Horn, the lead author on the paper and a Research Associate Professor of Psychological and Brain Sciences. "We wanted to advance and expand the cognitive neuroscience field by making the raw data accessible to more people for free."

According to the essay, the faculty who created the fMRI Data Center in 2000 were met with initial resistance from fellow neuroscientists. Some researchers were hesitant to give away their data; some questioned whether new science could arise from old data; and others thought that the technical hurdles could not be overcome. Despite these concerns, the Dartmouth group went ahead and teamed up with the Journal of Cognitive Neuroscience and made it a requirement to include raw fMRI data when submitting research to the publication. With initial support from the National Science Foundation, the W.M. Keck Foundation and the National Institute of Mental Health, the computer equipment was purchased and fMRI Data Center was established.

Now, four years later, the fMRI Data Center has archived more than 70 complete fMRI studies, which includes data from about 1,000 individuals. They have fulfilled, at no charge, more than 1,200 requests for data from researchers around the world. The overall size of their collection is about 2.6 terabytes (for comparison, about ten terabytes could hold the entire printed collection of the Library of Congress).

"We persevered in spite of the reception we received at first," says Van Horn, who serves as Operations Director for the fMRI Data Center. "We have created a repository of valuable information that continues to grow, and adding to it has become a natural part of the academic publishing process."

Van Horn wrote the piece with colleagues who were all instrumental in creating the fMRI Data Center: Scott Grafton, Professor of Psychological and Brain Sciences and the Director of the Dartmouth Brain Imaging Center; Daniel Rockmore, Professor of Mathematics and Computer Science; and Michael Gazzaniga, Professor of Psychological and Brain Sciences, Director of the Center for Cognitive Neuroscience and Dean of the Faculty. The authors note that early fears about fMRI sharing dat a have been overcome, and reusing data has proven fruitful to reanalyze previous findings or pursue related research questions that weren't considered originally. Using archived data for new studies provides new perspectives on data, the authors say, which contributes to the effort to better understand the fundamentals of human cognition.

The Nature Neuroscience essay is part of a special issue dedicated to research efforts that use enormous amounts of data and computational power, such as fMRI studies, proteomics and genomics endeavors that involve protein and gene analysis.