Right, I found a mask called "networks.nii", which allows me to delineate the 32 ROIs and now I am able to import the mask by selecting the "other clusters of interest (select mask/ROI)" in the REX GUI option available in the seed-to-voxel tab's Results explorer. However, I am still missing the beta values for the actual ROIs - I can see them in the Analysis Results section in the ROI-to-ROI tab. However, even though I can export the table that contains them in a .mat format, which can be accepted by REX, when I choose the .mat file after selecting the "other connectivity options (select second-level SPM.mat file)" option, REX does not accept it and gives a warning tone. Is there a workaround? 

Alternatively, couldn't I just stick with the connectivity values computed by the seed-to-voxel analysis but still use the networks.nii mask? Would there be a significant/important difference in the effect size?

Also, still it would be useful to know whether I can use some ROIs of the 32 by selecting them from the Results Explorer for the ROI-to-ROI tab and then exporting them, rather than using the entire set of 32 ROIs, mainly because I might want to only look at within networks. Or would that not affect the results and I can just focus on the ROIs that are within a given network after having selected a seed within this network? Probably?

Thanks,
Tsvet