Dear Juraj,
I guess you will have to run two separate 2x2x3 ANOVAs with GLM flex, one focusing on fMRI and one focusing on VBM / grey matter. You could then check whether any of the significant clusters in fMRI are associated with changes in grey matter and vice versa. This is not identical to the BPM approach though.
Alternatively, you have to rely on those model options which are included in BPM and its extension BPMe. But as far as I remember they don't allow for more complex designs, whereas the model options in SPM don't allow for a voxel-by-voxel covariate.
Best,
Helmut
|