jason-avery Wrote:
-------------------------------------------------------
> Hi, I'm trying to figure out the answer to a
> couple of questions I've had for a while now. I've
> been analyzing data on the cortical surface using
> a standardized mesh of -ld (linear division) of
> 125.
> My first question is:
> What is the actual grid size of this standardized
> mesh, in nodes/mm^2 (or in some other more
> appropriate metric)?
For ico128 (which is very close to 125) I get, using the intermediate surface (average of pial and smoothwm (white)) for a typical participant:
mean 0.596, std 0.303 (in mm^2 and mm, respectively)
the average edge length is 0.884 with std 0.324.
You can compute these (and other) statistics easily using the surf module in the latest PyMVPA (github.com/PyMVPA/PyMVPA), 'mvpa2.support.nibabel.surf.
> My second question is:
> I'm assuming from previous message board posts
> that higher mesh densities (like -ld 141, which is
> one of the default outputs of @SUMA_make_spec_FS)
> more closely approximate the original mesh density
> of most input cortical surfaces. If this is the
> case, then if we transform and analyze typical epi
> data on a cortical surface with this mesh density,
> are we essentially resampling (and possibly
> oversampling) the epi data (typically with
> original grid size of 2x2x3 mm^3 or higher) to the
> anatomical resolution (which is typically around
> 1x1x1 mm^3)?
Indeed you would be oversampling when using such a high-resolution mesh for EPI data.
But note that a high-res mesh also has the potential advantage of a more accurate selection of voxels within the grey matter, of course assuming good alignment between surface and EPI volumes.
Also, high-res meshes look prettier.