In the context of Skull-Stripping and Tissue classification. We've realized that results were better using a group specific template.

To create a group specific template, you can use DARTEL algorithm from spm.

First step is to use the New Segment functionnality for each subject you want to include into the template. Save Dartel Imported Images and Bias Field Corrected Images.

Once the first step is completed and you have checked the quality of the results, Use the DARTEL toolbox.

You have to run the DARTEL 6 times to produce 6 probabilistic template (GM, WM, CSF, Other1, Other2 and Background).

To create GM template, you have to use the image produced by the New Segment and starting with rc1.

To create WM template, you have to use the image produced by the New Segment and starting with rc2.

To create CSF template, you have to use the image produced by the New Segment and starting with rc3.

To create Other1 template, you have to use the image produced by the New Segment and starting with rc4.

To create Other2 template, you have to use the image produced by the New Segment and starting with rc5.

To create Background template, you have to use the image produced by the New Segment and starting with rc6.

DARTEL produce output called template_X.nii. BE CAREFULL !!! Change the name of the template to ensure that you are not overwriting it!!

Once Template are created. If you want to put it back into patient space:

1. Use Deformation field from DARTEL toolbox: Create Inverse Warp => Put it into DARTEL imported Space

2. Use Identity transform to put it into subject space.