Account for bad channels before GEDAI?

[GWeindel]

This looks very interesting, the theoretical as well as the examples are very convincing and I would like to give it a try. One question is regarding bad channel detections, is it useful/wise to apply an automated bad channel interpolation method before GEDAI or will it create problems (e.g. rank)?
Thanks!
Gabriel

[neurotuning-personal]

Hi Gabriel, Yes you can perform bad channel rejection in the same way that it is done before e.g. PCA or ICA. Although GEVD is actually quite robust to rank reduction, usually you would interpolate the channel after the ICA. So you can similarly run GEDAI on the channel-reduced data and then interpolate. However GEDAI does its own intrinsic interpolation in some sense (based on the theoretical BEM) which we have found to perform comparably to the classic spherical interpolation. That's why the GEDAI pipeline in the preprint simulated benchmark does not have a bad channel rejection step, whereas we included it for ASR/IClabel/MARA. I hope that makes sense. Cheers, Tomas ▬▬▬ *Tomas Ros, PhD* Lecturer, Department of Clinical Neurosciences CIBM EEG HUG-UNIGE Section University of Geneva, Switzerland https://www.tomasros.com/ [image: CIBM logo] *Clinical and Translational Neuroimaging* UNIGE Campus Biotech 9 Chemin des Mines CH - 1202 Genève, Switzerland (+41) 0766 158 863 cibm.ch <http://www.cibm.ch/> *Follow us: *[image: Twitter] <https://twitter.com/CIBM_ch> [image: LinkedIn] <https://www.linkedin.com/company/cibm-center-for-biomedical-imaging/> [image: YouTube] <https://www.youtube.com/channel/UCA8rMvimtkVzFwWzDDlaGMQ>

…

On Fri, 20 Mar 2026, 19:43 GWeindel, ***@***.***> wrote: This looks very interesting, the theoretical as well as the examples are very convincing and I would like to give it a try. One question is regarding bad channel detections, is it useful/wise to apply an automated bad channel interpolation method before GEDAI or will it create problems (e.g. rank)? Thanks! Gabriel — Reply to this email directly, view it on GitHub <#14?email_source=notifications&email_token=BOI5CW77QCOEIPKGKPE5RJT4RWGK7A5CNFSNUABAM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63RPHE3TAMRQGIZKM4TFMFZW63VKON2WE43DOJUWEZLEUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BOI5CW4AVNKL2GLEONEQVGT4RWGK7AVCNFSM6AAAAACWZOXIFWVHI2DSMVQWIX3LMV43ERDJONRXK43TNFXW4OZZG4YDEMBSGI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[GWeindel]

Hi Thomas,

That's terrific, it's now running without a bad channel detection step. Thanks!

Cheers,
Gabriel