An assessment of functional-anatomical variability in neuroimaging studies.

A key issue in functional neuroimaging is the amount of variability produced by individual differences in anatomical and functional patterns of activation. This variability affects summed images created when responses are averaged across subjects as well as comparisons between groups of subjects.In this report, functional-anatomical variability was explored at two different levels. The first level addressed whether responses defined in one group of subjects would replicate in a second subject group performing the same tasks. The likelihood that significant changes would be found in the second subject group was well-predicted by magnitudes and t-values of the responses in the first group.The second level of analysis addressed how closely the peak locations of changes in blood flow clustered together across subjects. The variability (mean vector distance) of peak locations among individual difference images was approximately 11.5 mm from the averaged peak location found across subjects. This value probably represents an upper bound for functional-anatomical variability using current PET data analysis techniques. Moreover, the variability was similar for responses distributed across different cortical areas and the cerebellum. This result is inconsistent with the hypothesis that some areas of the brain may have particularly high anatomical variability in normal right-handed subjects, thus precluding the use of averaging techniques for these areas. Copyright (c) 1996 Wiley-Liss, Inc.