Can you use solely EEG data to create an SSP filter?

  • MNE version: 1.2
  • operating system: Windows 10

Hello everyone!

I am currently working on a data analysis pipeline using solely EEG data that was recorded using NetStation. Because we used NetStation to acquire this data we only have EEG data to work with instead of MEG data and I do not have channels specifically dedicated to ecg recordings in the pipeline (yet i am hoping you all can help me with this). I am looking to implement an SSP filter to my data to repair EOG artifacts and I have noticed while reading the tutorials on MNE tools and watching other tutorial videos that most people create SSPs using EOG & ECG projectors and environmental noise reduction from empty-room recordings. I’m curious to know if you can create an ECG channel using solely EEG data so I can then create the ECG projectors and implement the SSP filter. Please let me know if you have any possible solutions to this problem or know other ways I could do this!

Hello @EmilyAJohnson and welcome to the forum!

It’s entirely possible to create EOG projectors from EEG data if you have some electrodes that capture ocular movements, like Fp1 or Fp2. Simply pass those channel names to compute_proj_eog() via the ch_name parameter.

Or construct the EOG epochs via create_eog_epochs() and then pass them to compute_proj_epochs() to create the projection vector(s).

For ECG, I’d say you’re out of luck.

Best wishes,
Richard

…because the cardiac signal is typically hard/impossible to recover from EEG sensors (whereas MEG magnetometers usually pick it up quite clearly). On the plus side, this means that including an ECG projector is not so important if all you have is EEG data, because they are usually less affected by it than MEG sensors. See, for example, this image https://mne.tools/stable/_images/sphx_glr_50_artifact_correction_ssp_009.png from this tutorial Repairing artifacts with SSP — MNE 1.2.1 documentation, where the EEG (black) lines clearly show blinks but don’t show heartbeats.

1 Like