SurfFWHM ACF Estimation?

I have a question regarding surface cluster correction in response to the Eklund FPR paper.

In volume space, I see that the recommended approach is to use the -acf option in 3dFWHMx to estimate noise smoothness; however, I can’t find a similar option for the SurfFWHM command.

As of now I’m using what I believe is he ‘classic’ FWHM estimate from the geometric mean of the SurfFWHM output. The surface blur estimates are close to what I’d expect with my blur level of 8, but the volume estimates are higher (~10-11) with the old-style parameters and the ACF model comes out around ~16-17.

Should I use the volume ACF estimate for surfaces? Or is there a way to estimate ACF for surfaces? Are surfaces as susceptible to this issue?

Thank you!


Hi Dan,

We have no surface method for ACF right now.

Note that when applying a FWHM blur on the surface,
SurfSmooth blurs to the desired level, whereas in
the volume, 3dmerge adds the requested smoothness.
So in the volume, the final estimates will often be
a voxel or so higher than the applied blur.

But yes, the ACF FWHM estimates tend to be higher

  • rick