overall your procedure is ok.
Still the interleaved comments down here.
Le 13/03/2011 2:24, Ben Becker a écrit :
> Dear SPMers,
> I’d like to preprocess my fMRI event related data according to the example data set in the spm8 manual (face fmri data / chapter 29). To fasten up the analysis I want to batch the preprocessing steps. Before running the batch I’d like to hear your opinion on some issues.
> My data:
> 2 runs of event related data (the same paradigm acquired twice, with a break of 1 minute, subjects stayed in the scanner)
> T1 structural image
> My preprocessing steps& my questions:
> 1. Realignment: Estimate& Reslice
> - Realign the 2 runs together (2 runs = 2 sessions)?
Yes, your "run" is what SPM calls a "session" (could be a 1min or 1month
break between them).
You only need to generate the "mean image", as the header of the fMRI's
will be adjusted with the estimated realignment parameters.
> 2. Slice timing
> - Correct the 2 runs in separate slice timing procedures or correct the 2 runs together (2 runs = 2 sessions)?
> - Data> session - Dependency: Estimate& reslice: resliced images?
No need to do slice timing.
Better use a more flexible GLM, i.e. if timing of activation is of
interest to you, include the derivatives of the hrf.
> 3. Coregistration: Estimate
> - Data> Reference Image - Dependency: Realign: Estimate& reslice: Mean image?
Yes, reference is the "mean image" produced at the realignment step (via
dependency). And as source select the structural image.
> 4. Segment
> - GM& WM output files in native space?
> - Data – Dependency: Coreg: Estimate: Coregistered Images?
Yes, use the coregistered structural image (via dependency).
GM & WM are not very useful (except as a visual check that segmentation
worked smoothly) but the bias corrected structural image is.
> 5. Normalise: Write
> - Data> Subject> Parameter file – Dependency: Segment: Norm Params Subj -> MNI?
> - Data> Subject> Images to write – Dependency: Slice timing Corr. Images (if slice timing for the two runs is separated in two
> separate slice timing procedures: select output from both)?
Use 2 "Normalise: Write" modules, for the fMRI and structural image
For the fMRI use the output of the realignement step and specify your
fMRI voxel size, for example 3x3x3
For the structural image, use a better resolution like 1x1x1 and select,
by dependecy, the bias corrected image from the segmentation step.
> 6. Smooth
> - Images to smooth – Dependency: Normalise: Write Normalised Images (Subj. 1)?
> Thanks in advance& kind regards,