Dear Thomas, > >thank you for your answer. I see that my problem could be that the >activated areas have a large magnitude and spatial extent and therefore >influence the average across all voxel values. >Maybe the scaling should not use the voxel values of all brain voxels >for computation of the scaling factor, but only voxels from CSF, where >we would not expect any useful signal changes. This would imply doing a >segmentation of CSF before scaling. > It would certainly be worth trying. However, when attempting to do that in the past I have often found that the CSF seems to have a life of its own, i.e. displays a variance, possibly from pulsatile flow/motion, not found elsewhere in the brain. Including this as a covariate would certainly "remove" the ventricles from your SPMs, but I am uncertain whether it would model "global" changes found in gray matter very well. Good luck Jesper Jesper Andersson Wellcome Dept. of Cognitive Neurology 12 Queen Square London WC1N 3BG phone: 44 171 833 7484 fax: 44 171 813 1420 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%