I have two groups of aged subjects, one normal control and one patient group, each n=10. I want to know (1) task-related activation in control group, (2) task-related activation in patient group, and (3) between-group differences in task-related activation. To enhance normalization, I decide to run DARTEL. I understand that in order to normalize EPI into MNI space, I need to run DARTEL to produce template_6.nii and flow field image (uxxx).
For this step, should I (A) put all 20 images in, and use the resulting template_6.nii and flow fields image for all subjects' normalization, or (B) run one DARTEL for controls and another DARTEL for patients, and then do normalization separately within each group?
If I use (A), it seems ok for (3) but problematic for (1) and (2)? (B) is ok for (1) and (2) but seems problematic for (3)?