This may be a silly question but, I am wondering if it is possible to parcellate functional data into an atlas in MNI space without having anatomical data to co-register the functional data to. Due to data-sharing issues, I have access only to preprocessed functional data. Is there a way to correctly align the atlas (in MNI space) and the functional data (in original subject space) without coregistering the functional scans to anatomical scans?
Due to brief Message Board outage, this issue was replied to by separate correspondence, but I'm now adding the reply here.
Hmmm, strikes me as quite tricky. Because the EPI is low-res and relatively low-contrast, typically, it really helps to have EPI-anatomical and anatomical-template alignments done separately and then concatenated. You can align the EPI to the template, but you basically have to expect that the anatomical fidelity will be lower (and you have to remember to use an appropriate cost function, like lpc, assuming the template has T1w-like tissue contrast).
Was your EPI blurred during processing? That will further reduce the quality of alignment. And also that would reduce your ability to average within ROIs and do cross-ROI connectivity, because the signal smearing would spread information artificially across ROIs. Actually, you can look at imperfect alignment as doing the same thing---ROIs will be "contaminated" by signal from separate ROIs, diluting signal within an ROI and artificially enhancing similarity (correlation) of signals across neighboring ROIs.
So, as you might have guessed, this seems like a difficult task. It can be done, but there will likely be a lot of caveats and concerns with interpreting the final results, in my opinion. Sorry to not have a better suggestion---inheriting data and using it for separate purposes is often challenging, because processing steps are often chosen for consistency with a particular goal in mind.
--pt