Hi

On 03/04/2016 07:06 PM, Bruce Fischl wrote:

I think there are some tools around for downsampling the surfaces like mris_decimate. Have you tried any of them?

Yes, indeed mris_decimate works well for the downsampling.

However, I'd like to recover the cortical parcellation for the downsampled surface. Is there a shortcut to do this, or does it need to go through the same steps of inflation, registration and parcellation as the original surface?

Thanks for answering my somewhat naive questions..

Marmaduke

what are you trying to achieve? It might be easiest to use something like fsaverage5 as the downsampled surface. Then you would just map your subjects to fsaverage5 using the existing sphere.reg files

cheers Bruce

On Fri, 4 Mar 2016, marmaduke woodman wrote:

On 03/04/2016 09:10 PM, marmaduke woodman wrote:

...

However, I'd like to recover the cortical parcellation for the downsampled surface.

I tried to decimate the pial & white surfaces, generate a corresponding thickness file and then

mris_sample_parc -ct /opt/freesurfer/FreeSurferColorLUT.txt tvb lh aparc+aseg.mgz decim.annot

but this seems not to generate a proper annot file. The same procedure works for the original pial & white surfaces.

The alternative, it appears to me, would be to generate decimated white, smoothwm, inflated, sphere, curv & sulc surfaces, apply mris_register to obtain decimated sphere.reg and then mris_ca_label to obtain a proper annot file for the decimation.

I will try this, but any insight would be appreciated.

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

I think you can map it to fsaverage5 then use mri_surf2surf with -sval-xyz to map the fsaverage5 surface back to each subject On Fri, 4 Mar 2016, marmaduke woodman wrote:

On 03/04/2016 10:06 PM, Bruce Fischl wrote:

...

what are you trying to achieve?

A per-subject ~10k vertex pial surface with a corresponding annot. We use it as a spatial domain for simulating neural activity; 150k vertices is too many for this purpose.

...

It might be easiest to use something like fsaverage5 as the downsampled surface. Then you would just map your subjects to fsaverage5 using the existing sphere.reg files

If I understand correctly, different subjects would then have identical cortical geometry, which isn't sufficient for our use case, as we are calculating the forward solution for EEG/MEG which can depend on the individual's cortical geometry.

Sorry if I've misunderstood your suggestion, I'm still a freesurfer noob.

Thanks, Marmaduke

---

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

On 03/04/2016 10:50 PM, marmaduke woodman wrote:

On 03/04/2016 10:19 PM, Bruce Fischl wrote:

...

I think you can map it to fsaverage5 then use mri_surf2surf with -sval-xyz to map the fsaverage5 surface back to each subject

Sorry for the noise, fwiw, this seems to do the trick:

mri_surf2surf --srcsubject tvb --trgsubject fsaverage5 --hemi lh --sval-xyz pial --tval pial.tvb --tval-xyz tvb/mri/T1.mgz

produces the 10k vertex surface in fsaverage5/surf/lh.pial.tvb, and

mri_surf2surf --srcsubject tvb --trgsubject fsaverage5 --hemi lh --sval-annot tvb/label/lh.aparc.annot --tval ./tvb-annot-on-fsaverage5.annot

produces a corresponding annotation.

thanks for your help! Marmaduke

Hi

The command

mri_surf2surf --srcsubject tvb --trgsubject fsaverage5 --hemi lh --sval-xyz pial --tval pial.tvb --tval-xyz ./tvb/mri/T1.mgz

works on dev (mri_surf2surf.c,v 1.103 2015/11/05 22:07:33 greve), but on stable fails with

ERROR: Option ./tvb/mri/T1.mgz unknown

Is there a simple workaround or do I need to upgrade from stable to dev?

thanks, Marmaduke