|
Abstract:
|
Increased computational resources and the availability of large databases offer new possibilities for the validation of neuroimaging analysis methods. In this talk, I show in a set of examples how these new possibilities can be used for the validation and extension of popular tools for meta-analysis of fMRI data. In a first study we examined the possibility to assess publication bias in coordinate-based meta-analysis. We simulated data-sets consisting of studies with either real activation or no activation at a predefined location in the brain. We investigated how the number of studies without activation in this location of interest that could be added to the meta-analysis before the congruency of activation was no longer statistically significant varied as a function of different multiple testing corrections and sample sizes. In a second study we used the single subject data of the IMAGEN database, a large multicenter genetic-neuroimaging study (Schumann, et al. 2010) to evaluate the effect of different group-level analysis models on the power, false positive rate and topological stability of coordinate-based meta-analysis.
|
