Hi Christine,

I'm not certain you'd want to smooth on the volume at all.

See this paper for illustrations
Desai R, Liebenthal E, Possing ET, Waldron E, Binder JR. (2005) Volumetric vs. surface-based alignment for localization of auditory cortex activation. Neuroimage 26:4 1019–1029.[CrossRef] [ISI][Medline]

Also, when smoothing on the surface (and clustering) in FreeSurfer, consider  that Freesurfer currently defines vertexes as part of the same cluster only when those vertexes are strictly adjacent (share an edge; there is no adjacency parameter that can be dynamically set), so when smoothing on the surface, you might want to use a decent smoothing kernel (If you use permutations methods to cluster, this would be controlled for as permutations control for smoothing).

U.

On 5/9/07, oxro03@yahoo.com <oxro03@yahoo.com> wrote:
Thank you very much!

I assume that when doing the first-level analysis with Feat and then project onto the surface, one shouldn't smooth so much in 3D volume space, right? How much smoothing should be used?

Thanks a lot,

Christine


Bruce Fischl < fischl@nmr.mgh.harvard.edu> wrote:
Hi Christine,

we have a set of tools we use mainly written by Doug called FS-FAST, but we
have also worked hard with FMRIB to make it easy to do surface-based
analysis using FSL. I think there are some tutorials on this on our wiki,
and perhaps on one of the Oxford sites as well (Steve and Doug: can you
chime in?).

Surface-based as a lot of advantages, mainly in terms of the accuracy of
cross-subject registration and the power gained in smoothing on the surface
as opposed to the volume. The disadvantages mainly come from the
possibility of misregistration and thus missing data, as you note. And of
course it doesn't represent activation in deep structures (e.g. caudate).

cheers,
Bruce


On Tue, 8 May 2007,
oxro03@yahoo.com wrote:

> Hi Doug & Ohers,
>
> I see that there are nice interfaces between Feat and Freesurfer for rendering Feat output onto the surface. Now for group fMRI stats it would obviously be advantageous to do all the stats directly on the surface, avoiding volume space altogether. Especially clsuter thresholding might be an issue here. Am I correct there?
>
> What approach would you suggest in terms of what software to use for the fMRI analysis. In our lab we have used mainly FSL tools to do volumetric fMRI analysis, and are hence most familiar with these, but are considering to moving to doing all this on the surface. What are people's views on that? Is that an entirely good thing or are there advantages to volume space analysis? What if spatial distortions are present in the BOLD images and haven't been corrected for? Any comments and suggestions on these issues would be greatly appreciated.
>
> Thanks a lot,
>
> Christine
>
>
> ---------------------------------
> Looking for earth-friendly autos?
> Browse Top Cars by "Green Rating" at Yahoo! Autos' Green Center.

Ahhh...imagining that irresistible "new car" smell?
Check out new cars at Yahoo! Autos.


_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer