Dear Johan,
Note that the "official" release of CAT12 / beta version was in March (at least the announcement on SPM mailing list), and it focuses on structural analyses. Thus not everyone interested in a better EPI normalisation might have realized that the toolbox comes with a Dartel template, and/or it might take a while till corresponding papers are published. Then there might also well be people who don't want to use that template for whatever reason, e.g. because CAT12 is still beta, because it is not "established" so far, ... In any case, previously there was just nothing like an "already existing" Dartel template (at least as far as I know), you would have had to create it from your own data. This might be one reason why Dartel has not been used that often when it comes to EPI data in general. fMRI studies are often based on a lower number of subjects, while it has been suggested that you should have a reasonably large number when trying to create your own templates, so people might have concluded that it is no option for their functional data.
With regard to preprocessing, make sure whether you work with "Segment" (as provided with SPM) or with "CAT12: Segmentation". In the latter, you can already select the existing Dartel template; after this step (during which you obtain the necessary flow fields) you have to normalise your EPI into MNI space still, for which you can choose among the following:
- via "Normalise: Write", based on the y_ flow fields
- via "CAT12: Apply deformations", based on the y_ flow fields (less settings that can be specified, e.g. no voxel size, bounding box)
- go with Util \ Deformations, based on the y_ flow field and a "pullback normalisation"
Finally, in theory there would also be the option to go with Util \ Deformations, based on the iy_ flow field and a "pushforward normalisation", however, this doesn't work with CAT12 iy_ flow fields, as they seem to represent some initial inverse flow fields and not the final ones (in contrast to those obtained with "Segment", here iy_ and pushforward should work well).
Note that the IXI Dartel template brain (but also the tissue probability maps included in SPM) is smaller than the latest MNI ICBM 152 non-linear atlas http://www.bic.mni.mcgill.ca/ServicesAtlases/ICBM152NLin2009 , so if you want your files to represent this MNI version you will probably need an additional step. In fact, there's the Dartel \ Population to ICBM Registration for exactly that purpose, so you could select the IXI Dartel Template_6_IXI555_MNI152.nii, then obtain another y_ flow field. When going with "pushforward normalisation" via the Util \ Deformations (note that you have to go with pushforward, not pullback, as the "y_" and not "iy_" might indicate), then you have an IXI Dartel Template corresponding to MNI ICBM 152 non-linear space. Maybe this is something for the CAT12 authors to consider, is it better to apply these normalisation parameters at the end of the preprocessing stream or would it be better to apply these onto the templates (possibly accomplishing the initial registrations then, as the different iteration templates would have to be transformed accordingly).
Best
Helmut
|