Hi, 

Thank you very much for replying my question last time. I really appreciate it. Today I have questions about seed to voxel analysis.

In seed-to-voxel analysis, how can I check the coordinates of seed voxel selected after the second-level analysis? 
In addition, is it possible for Seed-to-voxel analysis to make seed by meaning all voxel time series within ROI like ROI-to-ROI analysis? 

I also have another question about the spatial-resolution (voxel-level) in setup.
How can I determine the spatial resolution? Is it determined how I normalize structure and functional images?

For example, if I normalize structure image by defauly way (using T1.nii template implemented in SPM8 and voxel size 2x2x2mm), which should I select "same as template (2mm)" or "same as structure" as the spatial resolution in processing options setup ?

The reason why I'm confused it is that if the structure image is normalized in default normalization way in SPM8, the voxel size of the normalized structure image becomes 2x2x2mm,which means I can also choose "the same as structurals" in processing options setup. 

Since the results are influenced by spatial resolution, I'm concerning about it.  Thank you very much in advance.

Hi Peter,

In seed-to-voxel analysis, you define the seeds in the 'first-level analysis' step by selecting one or several 'sources' (these include all of your defined ROIs). Seed-to-voxel analyses compute first the mean timeseries within each ROI and then compute the correlation between these average timeseries and the BOLD timeseries at each voxel (after removal of confouding effects, filtering, etc.) If your original ROI consists of a single voxel then it will use that single voxel time-series as seed, while if your original ROI consists of multile voxels then it will use the average time-series as seed. Let me know if this clarifies.

And regarding the spatial resolution of the voxel-level analyses, you are right that if your structural volumes have been normalized (and resampled) using the standard SPM procedure then it does not matter whether you choose 'same as structurals' or 'same as template (2mm isotropic)' options in the 'spatial resolution' option (the results will be exactly the same). These different options are there to allow people with more complex sets of differing spatial resolutions in their data to choose the desired target spatial resolution of the voxel-level analyses (in particular it is not uncommon to have structural data resampled to 1mm isotropic voxels, while still preferring to perform the functional analyses using coarser resolutions -e.g. 2mm isotropic voxels, or the original functional volume resolution for subject-based analyses-)

Best
Alfonso

Originally posted by Peter Dessen:

Hi Alfonso, 

Thank you very much. I figure out the way of Seed-to-voxel analysis. In addition, your reply helps to make sure that there is not difference between "same as structure" and "same as template" if the structure data is normalized into SPM T1 template. I really appreciate it!

Kind regards, 
Peter

Originally posted by Alfonso Nieto-Castanon: