Abstract #1018

Pardoe H, Pell G, Abbott D, Jackson G, Briellmann R

Voxel-based morphometry (VBM) is a popular technique for the analysis of volume differences between groups of structural images. It is believed to be limited to comparisons of the images acquired with the acquisition scheme and on the same scanner. This study was undertaken to investigate the potential of multi-centre VBM (i.e., analysis of images obtained with different acquisition methods or at different sites) by investigating the sources of confounding variance that would be introduced. The effects of differing voxel sizes and orientations and that of different RF coils are shown to be especially significant, but pooling the data appears to diminish the likelihood of artefactual results.