REPRODUCIBILITY OF FMRI RESULTS ACROSS 4 INSTITUTIONS USING A SPATIALWORKING-MEMORY TASK

Four U.S. sites formed a consortium to conduct a multisite study of fM RI methods. The primary purpose of this consortium was to examine the reliability and reproducibility of fMRI results. FMRI data were collec ted on healthy adults during performance of a spatial working memory t ask at four different institutions. Two sets of data from each institu tion were made available. First, data from two subjects were made avai lable from each site and were processed and analyzed as a pooled data set. Second, statistical maps from five to eight subjects per site wer e made available. These images were aligned in stereotactic space and common regions of activation were examined to address the reproducibil ity of fMRI results when both image acquisition and analysis vary as a function of site. Our grouped and individual data analyses showed rel iable patterns of activation in dorsolateral prefrontal cortex and pos terior parietal cortex during performance of the working memory task a cross all four sites. This multisite study, the first of its kind usin g fMRI data, demonstrates highly consistent findings across sites. (C) 1998 Academic Press.