Meta-analysis of functional imaging data using replicator dynamics

Abstract

Despite the rapidly growing number of meta-analyses in functional neuroimaging, the field lacks formal mathematical tools for the quantitative and qualitative evaluation of meta-analytic data. We propose to use replicator dynamics in the meta-analysis of functional imaging data to address an important aspect of neuroimaging research, the search for functional networks of cortical areas that underlie a specific cognitive task. The replicator process requires as input only a list of activation locations, and it results in a network of locations that jointly show significant activation in most studies included in the meta-analysis. These locations are likely to play a critical role in solving the investigated cognitive task. Our method was applied to a meta-analysis of the Stroop interference task using data provided by the publicly accessible database BrainMap DBJ.

Figure 1

Co‐occurrences of activation foci from Table I represented as a graph. Nodes represent activation foci and weighted connections represent the number of co‐occurrences of two foci.

Figure 2

Top: 239 activation maxima obtained for fMRI experiments addressing the Stroop interference task. Bottom: 15 activation foci derived from ALE of the above coordinates. Activation maxima and foci are shown in the glass brain view, i.e., projected onto three orthogonal single coronal, sagittal, and axial slices.

Figure 3

The five selected members of the dominant network. They include the IFJ, IFS extending into middle frontal gyrus, pre‐SMA, and ACC of both hemispheres extending into BA8. a, b: Left lateral cortex at x = −44 and x = −3, respectively. c: Right lateral cortex at x = 7.