Hello SPM listserv,
My colleague and I were working on some resting state analyses and employing DARTEL for normalization. An initial pass through, following the pipeline outlined in the SPM manual, yielded very mediocre results. We were seeking advice about improving our results (or if there was something we were potentially doing wrong).
Our pipeline included (in SPM12):
-Registering (but no-reslice) coplanar T1-weighted scans to our EPI volumes (EPIs were the reference)
-Normalizing the GM and WM from our coplanar T1-weighted images to 6 DARTEL templates (from the VBM toolbox)
-Applying the flow fields to the EPI volumes
We have checked the registration between the coplanar T1-weighted scans and EPIs (and they seem fine). We have also checked the registration between coplanar T1-weighted scans and the DARTEL template (and again, fine). Normalizing the EPIs using the DARTEL flow fields however seem off. Basic visual inspection would suggest that the DARTEL flow fields fared no better (perhaps worse) than typical affine registration.
A few example subjects are linked below:
http://jamiehanson.org/SPM_list/20120207_14524_swuaV0001.img
http://jamiehanson.org/SPM_list/20121025_16249_swuaV0001.img
http://jamiehanson.org/SPM_list/20121129_16475_swuaV0001.img
(hdrs also uploaded to same addresses, simply change link to .hdr)
We have also altered the pipeline to register (but no-reslice) the EPI (with coplanar T1-weighted scans as our reference), and the output looks even worse. We wondered if others had experienced this issue, or what things we might need to troubleshoot to improve our results. Any suggestions or thoughts are greatly appreciated!
Best,
Jamie.
|