Hello Jiang, Marco, others, the statistical values one sees at p<0.001 indeed seem as if caused by factors 'unrelated to the disease' but technical/model-related. - Before changing the model, all modulated and modulated/smoothed images should be checked one by one to exclude any 'major accidents'. - total gray matter as covariate - absolutely agreed and surely necessary; as the affine transformations translate into each voxel value, each voxel value (in turn) contains information on total gray matter. - total grey matter captures most aging effects and global disease effects. So, best also look at total grey matter values (better: grey matter divided by total intracranial volume) in an extra (offline) analysis. If the groups are not age matched (as they are not equally sized, they can't be on a case by case basis...), age may be included or at least probewise be included. - Was "cleanage" of segmentated imges switched on? There seems some dural tissue left in the midline which usually is cleaned away. - And: keep us updated... best regards, Philipp Max Planck Institute of Psychiatry NMR Research Group Kraepelinsr. 2-10 80804 Munich Mail: [log in to unmask] Phone: 0049-89-30622-413