Is it possible to manually define brainstem ROI in MEG volumetric source analysis?

Support & Discussions
1

Hi all,

I’m working on a project aiming to localize brainstem responses using MEG. I’m currently using fsaverage with the aparc+aseg atlas for volumetric source reconstruction.

I noticed that the brainstem is defined as a relatively large ROI in this atlas, which may be too coarse for my purposes. I’m wondering whether there are recommended approaches to creating a custom atlas (e.g., modifying aparc+aseg) to subdivide the brainstem into smaller ROIs.

More specifically:

  • Is it possible to manually define or draw finer parcellations for subcortical structures like the brainstem?

  • Are there existing atlases or workflows compatible with MNE that provide more detailed brainstem segmentation?

I’m aware that subcortical localization is challenging with MEG, but I’d like to explore whether finer anatomical constraints are feasible. Any suggestions or pointers would be greatly appreciated!

2

There are several subcortical atlases that are compatible with freesurfer and then also MNE (with some extra work). Some are incorporated with freesurfer and other require extra setup (the AANSegment requires an external matlab runtime, but then looks like it outputs a freesurfer compatible output). You have already mentioned that these are going to be difficult to localize with MEG, so it could be a lot of extra work with little payoff. Below are several links to additional subcortical segmentations:

-Jeff

3

Hi Jeff,

Thank you for the suggestion. I received a solution from Eugenio, the original author of the brainstem substructures. I have now obtained the brainstem atlas for the fsaverage template brain.

Zoe