Joe Devlin is right, of course, there are characteristic distortions in the EPI data set that could cause some trouble.
For what it's worth, we use an atlas derived from registered and averaged high resolution T2-wtd EPI scans, per the work of Woods, Dapretto et al. HBM 1999;8(2-3):73-9. This atlas is in Talairach space and is an efficient registration target from EPI data sets - more reliable than the avg152T1 that comes with the FSL distribution, because it has contrast behavior more like the functional data sets.
You can find the atlas images at: http://airto.loni.ucla.edu/BMCweb/HowTo/RefImages/
|