Hi,
We're hoping to use 3dClustSim to define the cluster extent (# of voxels) needed to reach significance for VBM data, which we analyzed with the VBM8 toolbox in SPM8. We'll need to obtain smoothness estimates from 3dFWHMx.
Since the instructions 3dFWHMx are geared towards fMRI data and suggest using the errts, it is not clear what to use to calculate smoothness estimates from structural data in VBM analysis. Since we do not run individual level models for VBM (these are 3D instead of 4D datasets and we don't remove motion/conduct statistics at the indivdual level), we don't have errts (residuals) at the individual-level.
Specific questions:
1. What should be used as the input to 3dFWHMx for VBM data?
Would it be the smoothed gray matter images (which are the inputs to our group-level statistical models)?
e.g., simply "3dFWHMx -automask -input smoothed_GM_$subj" or are more options needed?
2a. Once we have the fwhm values, are there any other options we should specify in our 3dClustSim command given this is for structural VBM data? e.g., is it simply "3dClustSim -dxyz 1.5 1.5 1.5 -fwhmxyz sx sy sz -prefix output"? Are there other options we need?
2b. Does it matter if we specify "-dxyz 1.5 1.5 1.5" (VBM8 toolbox automatically resamples to 1.5x1.5x1.5) or use the "-mask" option?
With the "-mask" option, would it be appropriate to use our custom template gray matter image, or should this be some other mask?
Thanks,
Andrea