> the reason Bet performs badly is that it is a whole head scan, so the > initial Bet sphere is not contained within the brain. > If you take an ROI which has the brain roughly in the middle > avwroi 3d 3d_roi 0 170 0 256 100 150 > then run Bet on this > bet 3d_roi 3d_roi_brain -f 0.5 -g 0, > The results are pretty good. Indeed they are! Thank you very much. Could I suggest that this remark is added to the manual? Thanks, Erik-Jan Vlieger