CSSP and other advanced algorithms for BCI decoding

💬 Support & Discussions
1

Dear all,

As you know, BCI literature has been booming for 2 decades, and new algorithms have been developing. Currently, MNE has only CSP algorithm as a spatial feature extractor for BCI decoding. However, it would be amazing to add CSSP (Common spatio-spectral pattern) algorithm to MNE. I am a newbie here so I could contribute at some point but CSP.py is already pretty complicated for me. Is someone planning to add CSSP & RCSSP or some other algorithms for BCI decoding?

Here are the papers:

  • CSSP: “Lemm et. al, 2005: Spatio-Spectral Filters for Improving the Classification of Single Trial EEG”
  • RCSSP: “Cherloo et. al, 2021: Ensemble Regularized Common Spatio-Spectral Pattern (ensemble RCSSP) model for motor imagery-based EEG signal classification”
  • SBCSP: "Novi et. al, 2007: “Sub-bandCommonSpatialPattern (SBCSP) for Brain-Computer Interface”

Cheers

2

are you aware of existing Python implementations of these methods?

Alex

3

I haven’t found any of them.

4

then I think it’s the starting point. Draft a working implementation and see how it goes

maybe https://pyriemann.readthedocs.io/en/latest/ has some alternative already.

Alex