Volume based Destrieux atlas

[gianmarcoduma]

Hi everyone, I'm trying to implement a dynamic functional connectivity clustering method from fMRI to EEG. I'm computing connectivity in brainstorm and instead and I'm using the destrieux atlas. Unfortunately I need to convert the Destrieux from a surface based to a volume based atlas into MNI space with 2mm voxels. Is there a process to do it in brainstorm, or do you have any suggestion to do it?
thanks

[Francois]

Sorry for the response delay.

No, there is no easy solution to convert between surface and volume atlases.
But, there are already available in FreeSurfer. After processing your MRI with FreeSurfer, you have access to the Desikan-Killiany (aparc+aseg.mgz) and Destrieux (aparc.a2009s+aseg.mgz) atlases as volumes:

Freesurfer template aparc+aseg vs aseg atlas

The default Brainstorm labelled aseg atlas (tess_aseg.mat) is a pleasure to use, however I need also the bilateral subcortical region "ventral DC" (region #28 in the screenshot under section 2 here)... Is it at all possible to have a version with it included? (In other words a tess_aparc+aseg.mat?) Otherwise, is there some simple way to add it? Grateful for any help!

https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/AnatomicalROI

[gianmarcoduma]

Thanks!!

[gianmarcoduma]

Just a brief question: do you know where I can download the MNI template with the T1 with voxel resolution of 2mm instead of 1mm?
Thanks

[Francois]

Maybe an old version of SPM?
I wouldn't recommend this: this would mean an old version of the template. Probably either MNI305 or Colin27.
http://www.bic.mni.mcgill.ca/ServicesAtlases/MNI305
http://www.bic.mni.mcgill.ca/ServicesAtlases/Colin27

You could also compute a downsampled version of the 1x1x1mm voxel yourself.
Why would you need this?

[gianmarcoduma]

I'm sorry for my late answer.
I'm trying to implement a method to clusterize dynamic functional connectivity. One of the steps is to compare the clusterized activity with the Yeo networks. In order to do that I need to apply a correlation between the clusterized network computed from my Destrieux atlas and the Yeo network. I'm using a function from other researchers; this function use the volume based (2 mm voxel) Yeo networks, therefore I need to obtain the destriux atlas in the volume with voxel size of 2 mm, in order to run the correlation
I run recon-all on the MNI- 2 mm voxels, but when I import the .mgz Destrieux atlas I still have a matrix of 256x256x256 instead of 91x109x91 like a 2mm voxels MRI.
It's a bit tricky because I'm trying to implement a method from fMRI in EEG resting state. I don't know if it will work.

[Francois]

@gianmarcoduma

I run recon-all on the MNI- 2 mm voxels, but when I import the .mgz Destrieux atlas I still have a matrix of 256x256x256 instead of 91x109x91 like a 2mm voxels MRI.

Indeed, by default, FreeSurfer rewrites the input volume in a 256x256x256 volume with 1x1x1mm voxels. I think in FreeSurfer 6, recon-all has an option to keep the original volume size.

But the easier would be to first export your source volumes and then downsample them.
The process "File > Export to SPM8/SPM12" lets you downsample the output volume by an integer factor. Enter "2" and you get a 2x2x2mm voxel size in output.