 |
 |
Dear Jessica, > Is there a website on this method? :) I still have a few questions-- I guess, there will be one, but I don't know yet. > --- John Ashburner <[log in to unmask]> wrote: > > Just do pretty much what the Good et al paper suggests: > > > > 1) Coregister early and late scans together. > > 2) Segment one of these images. > > Should these images be already normalized to the SPM templates? (Does that > mess up the volume preservation procedure later?) > Or can SPM segment non-normalized images? This is the special trick in the VBM method, presented by Good et al. that you first segment the non-normalized images. SPM can segment non-normalized images by normalizing the images internally, segmenting them and then transforming these images back to the original space. But usually, especially in segmenting non-normalized images, you have also some voxels outside the brain. Therefore, the extract-brain option, described in John's previous mail, is rather important at that point. The idea behind this method is, to get a optimised transformation into the MNI space, if you normalize only the masked seg1-image onto a grey-matter template. So, you optimise your normalization by doing this especially for the grey matter (and white matter, if you do the same thing also with the seg2 image and a white matter template). After applying this transformation to the non-normalized, non-segmented original images, you have to segment these normalized images again. And these resulting seg* images are the ones, you will use in the VBM analysis. > > 3) Use the extract brain option to obtain a mask of brain tissue. > > 4) Use Imcalc, selecting *_seg1.img, *_seg2.img and brain_*.img > > in order to generate cleaned up grey matter images by entering > > an expression like i1./(i1+i2+eps).*i3 > > 5) Estimate spatial normalisation parameters from this cleaned up grey > > matter image by matching it to a grey matter template. > > Is it ok to use the apriori gray.img that comes with SPM99? The best way is to construct your own template, but, especially in a study with a growing number of investigated subjects, it should be okay, to use the grey and white matter templates from the apriori directory. I have rather good experiences by using that templates instead of creating an own one. ---------------------------------- Karsten Specht fMRI Section Department of Neuroradiology Medical Centre Bonn Spessartstrasse 9 53119 Bonn Germany Phone: ++49-(0)228/90 81-178 Fax: ++49-(0)228/90 81-190 E-Mail: [log in to unmask] WWW: http://www.mcbonn.de/Praxis/praxis15/fmri1.htm