Hi everyone,
I'm making a tool for easier finding fNIRS neural connectivity with Cedalion, and I need a good way to group channels.
I hoped this could be achieved using parcellation data, as imported with a head model (Colin 27 -> head_model.brain.vertices) and relating that to my channels via the Recording's geo3d object (-> channel midpoints) once aligned with the model, but I'm not having much luck.
Anything from advice to wholesale solution would be greatly appreciated, please let me know if there's any other info needed.
Thanks,
Adam
Hi Adam,
i assume you are referring to using parcellation labels but without having done image reconstruction before that, correct? In the case of image reco, you can check the example here to see how you can group vertices by parcels in image space.
If you look for a way of mapping from parcels to channels you can check out the example here that shows how to use the fwm.parcel_sensitivity function to map between parcels and channels.
Note that also for this case you will need to once run the forward model (photon simulation) to get the sensitivity profile between parcels and channels - you won't really be able to do without, because sensitivity depends on your channel distribution/coverage and distances. It is also not hard, though - how to run the forward model simulations you can find in the image reconstruction example notebook. Note that you do not need a high density fNIRS probe to do image recon / run the forward model, you can do that for any channel configuration.
Hope this helps,
Alex
Thanks for this Alex, it makes sense to me now how the sensitivity is connected and I've got it running now!
@aga165 just in case: the new 2026.05 release also comes with a head model example notebook that shows some more easy ways of using and visualizing parcels
https://doc.ibs.tu-berlin.de/cedalion/doc/dev/examples/head_models/43a_head_models_overview.html




