kblacker Wrote:
-------------------------------------------------------
> At
> this point, I have created surface models using FS
> and run @SUMA_Make_Spec_FS (which I understand
> from the handouts is the point at which my
> standard meshes are created, correct?). I have run
> preprocessing using afni_proc with the surf block
> included. I structured my afni_proc off of example
> #8 in the afni_proc help. Now, I want to run
> individual subject analysis using 3dDeconvolve (I
> did not originally include a regress block in my
> afni_proc) and then I will want to run group-level
> analysis. My question is: at what point do these
> standard meshes get used?

Without afni_proc.py, one would typically run preprocessing up to and including motion correction (3dvolreg), then map the functional data onto the surface using 3dVol2Surf, and then do the remaining steps (spatial smoothing, signal normalization, 3dDeconvolve) using the surface data.

I don't use afni_proc.py, but I assume it does the same thing. If you're not running 3dDeconvolve (or 3dREMLfit) using afni_proc.py, you could just feed the surface data files (.niml.dset) to 3dDeconvolve.

> Should I be mapping my
> individual level analysis onto my standard meshes?

It's better to do the mapping early (after motion correction) than late (after 3dDeconvolve).

> In order to do group-level analysis, I need all of
> my participants registered to a standard space,
> just like in volumetric analyses, correct?

Yes and no.

Yes: the standard space here is a common topology of the anatomical surface files (*.asc) that were used to map individual subject data. Typically one would use MapIcosahedron to get such surfaces.

No: the anatomical surface files for different participants can all be in their own native space with respect to the spatial coordinates of the vertices. It is not necessary, for example, that surface files from different participants share the same center of mass.

> How do
> I do this with the standard meshes?

Simply give the beta estimates in the surface data files (.niml.dset) from the 3dDeconvolve output to 3dttest++ (or another program you want to use for group analysis). It's very similar to what you would do with volumetric data.

Note that if you don't have full-cotex coverage, there may be nodes without data, which is something 3dttest++ et al don't like. To fix this, use ConvertDset with the -pad_to_node option.