template surface files

[nmarkowitz]

Hi,

I was wondering if there are surface files (such as giftis or freesurfer surfs) for the volumetric MNI templates. I've seen some in various packages but I haven't come across an archive that contains them. This would be very useful for plotting and projection. I work with ieeg data and surfaces are a pretty standard plotting method and assist in snapping electrodes to the cortical surface. The fsaverage and fsLR templates have surface files but I haven't seen them for others such as Colin27 or the various MNI templates. Is this something possible to do or already exists? One thought I've had is to use fmriprep's output that would place creates surfaces in the original volumetric space. Any input or advice is greatly appreciated.

[oesteban]

there are surface files (such as giftis or freesurfer surfs) for the volumetric MNI templates.

No, we don't have them. But I agree they would be useful.

I've seen some in various packages but I haven't come across an archive that contains them. This would be very useful for plotting and projection. I work with ieeg data and surfaces are a pretty standard plotting method and assist in snapping electrodes to the cortical surface. The fsaverage and fsLR templates have surface files but I haven't seen them for others such as Colin27 or the various MNI templates. Is this something possible to do or already exists?

Agreed. If you can point us to some sources where we can get these surfaces, and if they are licensed open, then of course this would be a more than desirable contribution to have.

One thought I've had is to use fmriprep's output that would place creates surfaces in the original volumetric space.

That's an option. I would look at Yeo's work before trying anythin.

[nmarkowitz]

I checked Yeo's lab's github repo and there is an MNI152 (not sure which version) and a Colin27 freesurfer reconstruction. This is useful but it's not all of them. I've found one other freesurfer reconstruction of an MNI152 template here.

As an additional question, I'm not as knowledgable about all the atlases, are all the MNI atlases in the same space? That is, if you register a volume to the MNI152Lin then should that same volume also be in MNi152nonlin_asym space? For freesurfer output, the dimensions of the volume changes (to 256x256x256). How do you correct for this? Is fsaverage aligned to MNI152 (they all, when having the same dimensions, are aligned by ACPC?). Sorry for all the questions.

While some work on this has already been done, it may be good to have a freesurfer recon of the same version run across the templates and move the surfaces into their native spaces. One nice thing that surface parcellations can be applied across all templates in their native spaces.

[oesteban]

Thanks for your patience @sportnoah14 -- I've been on vacation since Dec 2.

I checked Yeo's lab's github repo and there is an MNI152 (not sure which version) and a Colin27 freesurfer reconstruction. This is useful but it's not all of them. I've found one other freesurfer reconstruction of an MNI152 template here.

We don't have the Colin27 template on templateflow (I'd be happy to assist in adding it), which is the only one of those where it would make sense to run FS's recon-all.

The other two alternatives (FSL_MNI152_FS4.5.0/ at Yeo's lab repo) and @chrisgorgo's work with MNI152NLin2009cAsym are surfaces generated with recon-all run on the template - which I believe is not the best way of doing this.

Although we are light in resources, our idea for this was to average surfaces together from individual subjects instead. But that's definitely not going to be available anytime soon. I know Yeo has published several pieces on the problem.

As an additional question, I'm not as knowledgable about all the atlases, are all the MNI atlases in the same space? That is, if you register a volume to the MNI152Lin then should that same volume also be in MNi152nonlin_asym space? For freesurfer output, the dimensions of the volume changes (to 256x256x256). How do you correct for this? Is fsaverage aligned to MNI152 (they all, when having the same dimensions, are aligned by ACPC?). Sorry for all the questions.

No, they aren't all in the same "space". They are aligned on the same ACPC plane and are very close in scale (although you can clearly eyeball a little mismatch between MNI152NLin6Asym --"FSL MNI"-- and MNI152NLin2009cAsym). So they are quasi-aligned. fsaverage is a surface template, that is aligned --if I'm not wrong-- to a version of MNI305 translated ~4mm orthogonally to the axial plane (which is aligned with ACPC).

While some work on this has already been done, it may be good to have a freesurfer recon of the same version run across the templates and move the surfaces into their native spaces. One nice thing that surface parcellations can be applied across all templates in their native spaces.

As I introduced before, we have plans to register many subjects into the different MNI spaces and then project the corresponding surfaces to finally "average those surfaces" (see Fig 2 in the OHBM poster at the bottom of https://www.templateflow.org/).

[nmarkowitz]

@oesteban, thanks for the detailed response. I think those surfaces would be a great addition to the templateflow archive and I look forward to them. This entire thread has been very helpful and informative for me. I'll close this issue

[nmarkowitz]

I thought I would post this into this already existing thread. Freesurfer now has this new template called "cvs_avg35" and "cvs_avg35_inMNI152". The latter is in MNI152 space (here's a post briefly explaining how they were made). If I transform a volume to MNI152 space (such as the MNI152lin) then can I consider it to be in the same space as the cvs_avg35_inMNI152 freesurfer template? This could be a nice compromise to have a template with surface reconstruction in MNI152 space.