Dear Adeel,
Many thanks to your reply. Sorry for my incomplete description about the process.
For the preprocessing, I used AFNI to do all steps including discarding first 10 volumes, despiking, slice-timing correction, motion correction with six-parameter rigid body transformation, and then spatial normalization of the T1 images registered to MNI152 template and a 4mm FWHM smoothing. Later motion censoring, nuisance regression and bandpass filtering (0.009–0.08 Hz) were also done simultaneously. And no volumes were censored for this subject.(But what if some volumes(e.g. around 20 volumes) were censored? Will it have influence on following DCM analysis?)
I have no idea if the preprocessing steps will have such a huge influence on the fitting. I then used SPM to do the preprocessing for the same subject as well. The steps included realignment, coregistration, segmentation, normalization and smoothing (just following the manual preprocessing steps). Same GLM specification and DCM steps were done to the SPM-processed data. You can find the fitting results for this one in attachment. It looks better for some of the fittings than my previous one but still looks poor compared to what the manual shows. Is this one a "normal" fitting?
BTW, the 6 ROIs are primary motor area, putamen and cerebellum of both hemispheres with the radius of 6 mm for your reference.
Thanks very much for your time and looking forward to your reply.
Best regards,
Tracy
|