 |
 |
Dear Dr. Shroeder, This is just an additional help. Dr. Furman wrote: >A lot of settings for spatial normalization are hidden in the defaults >(Defaults button) and therefore they are not seen by users. You must >customize your defaults EVERY TIME you do spatial normalization - to >choose the correct model, voxel size, template, and STARTING ESTIMATES. If you feel it bothersome to change these parameters EVERY TIME you analyze your data, you can edit spm_defaults.m file in the spm96 directory using vi or some other editors and customize the defaults as you like. In your case, the following line (or somewhere like this line) should be edited: sptl_Ornt = [0 0 0 0 0 0 -1 1 1 0 0 0]; % Radiological Convention (L is R) I do not know exactly how you should change this line, but as far as I judge from your e-mail, you should change the seventh number (-1 replaced with 1, or 1 replaced with -1). As you see in the spm_defaults.m file, you can edit other default parameters such as number of parameters for affine normalization, number of nonlinear basis functions, etc. I hope this helps and you enjoy a confortable SPM life. Best wishes, Kota KATANODA Department of Cognitive Neuroscience, University of Tokyo, Japan e-mail:[log in to unmask] >> I'd like to know if anyone has had a problem with running a statistical >> analysis of spm95 processed (realigned, normalized, smoothed) images in >> the spm96 version. In comparison with the statistical maps generated >> with these images within spm95 it seems that the spm96 maps are showing >> activation which is laterally flipped (x = -x) but otherwise almost >> identical (i'm guessing they are not identical because of thresholding >> default differences between the two versions). >> > > >I believe the problem is in the different "conventions" used by SPM'95 and >SPM'96. The statistical output is always in "neurological" conventions >(image left = subject left) but SPM'95 is flexible on the input end (so >it asks you this question before doing statistics). If your image is >in Radiologicxal convention (image left = subject right), SPM'95 >will flip it. SPM'96 is not flexible - it expects ALL input in >neurological convention, and the flipping must >be done when you do spatial normalization. > >A lot of settings for spatial normalization are hidden in the defaults >(Defaults button) and therefore they are not seen by users. You must >customize your defaults EVERY TIME you do spatial normalization - to >choose the correct model, voxel size, template, and STARTING ESTIMATES. In >your case, your STARTING ESTIMATES should be Radiological convention (if >your images are acquired that way). > > >Take care, > > >Vitaly Furman >Brain Imaging Lab >Columbia-Presbyterian Medical Center >New York, NY %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%