you can use mris_fwhm with the mask option. Just give it a maks of the places that are NOT NaNs doug

On 05/21/2013 09:37 AM, Glen Lee wrote:

Inline image 1

Hi freesurfer experts,

I'm dealing with an issue with smoothing on surface data that
unfortunately contain 'NaN' in the ventral temporal region (which
is a region that we are interested in).

This may be due to the slab data acquisition, where the bottom of
the slice meant to be at the bottom of the tomporal lobe for every
subject. Apparently, we didn't quite succeed in this as we saw
missing data points that are filled with 'NaN' during 2D
reconstruction in some subjects' 2D data.

In the cartoon above, the yellow color depicts vertices with 'NaN'
in the ventral temporal region , and you can easily see how this
is affected by smoothing (4mm FWHM), such that a large swath of
ventral regions have NaNs after smoothing.

I've tried smoothing both with and without --cortex flag, but the
results were the same regardless.

So, the best bet would be masking out those NaN vertices before
smoothing and I wonder if there is any way of doing it in freesurfer.
Otherwise, I may need to consider smoothing outside of the
freesurfer using Matlab (but I have no experience about this
approach).

If anyone has previously dealt with this issue (hope I'm not the
only one) and know how to get around with it either inside or
outside of the freesurfer, please help me.
-Glen

---

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