 |
 |
Dear Fuzhenrong, I would suggest to turn to SPM12 and rely on the "Segment" module (this corresponds to the previously known "New Segment" but has been modified in some ways, also the manual has been updated). I'm not sure why you want to normalize the structural volumes into the space of the rc files. The rc files are required to create a study-specific template with Dartel, but then you would want to normalize the structural volumes onto that template as well. Thus usually you would go with the following steps (one step does not necessarily correspond to one "module" in SPM GUI window) 1) Segmentation 2) Create template 3) Normalize data to that template While creating the template you should also obtain files called u*, these files store the information necessary to normalise your data (e.g. the raw s*, the c1, c2) onto the template that you've just created. There should be several different options on how to normalize data then, but one option would be to go with the "Normalize to MNI Space". The name is misleading, but well. Now there are two options: 1) You want to normalize your data onto the created template with no other additional transformations. Click on "Dartel Template", but do NOT specify any image, just click on "Done". For "Select according to" specify your different subjects with the corresponding flow fields (e.g. u_rc1sANI001_nameofyourtemplate) and the data you want to normalize (e.g. c1sANI001 and sANI001). Adjust the other settings accordingly, also make sure whether you want to modulate the data or not (see "Preserve") and whether you want to apply smoothing (if you don't want to set "Gaussian FWHM" to [0 0 0]). 2) As your template differs to some extent from the initial template / TPMs used during segmentation, coordinates will not match exactly. If you want to make your template closer to the orginal template you can estimate *affine* transformations required to bring your template into the space of e.g. the SPMmouse template / TPMs. For that purpose you could specify one of these files as "Dartel Template". This will not decrease accuracy of your own template, it's just an additional transformation which would be applied to all your data to exactly the same extent (in a very simple case, maybe stretched by factor 1.5). Given that the initial template is probably of a rather low quality I would probably not do so, but it *might* be an option to consider. Instead of 1) it should also be possible to go with "Deformations Utility" or the "Normalize: Write" module or the "Dartel: Create Warped", you'll have different options with different modules. Best Helmut