Functional volumes modeling: theory and preliminary assessment.

A construct for metanalytic modeling of the functional organization of the human brain, termed functional volumes modeling (FVM), is presented and preliminarily tested. FVM uses the published literature to model brain functional areas as spatial probability distributions. The FVM statistical model estimates population variance (i.e., among individuals) from the variance observed among group-mean studies, these being the most prevalent type of study in the functional imaging literature. The FVM modeling strategy is tested by: (1) constructing an FVM of the mouth region of primary motor cortex using published, group-mean, functional imaging reports as input, and (2) comparing the confidence bounds predicted by that FVM with those observed in 10 normal subjects performing overt-speech tasks. The FVM model correctly predicted the mean location and spatial distribution of per-subject functional responses. FVM has a wide range of applications, including hypothesis testing for statistical parametric images. Copyright (c) 1997 Wiley-Liss, Inc.