Iām working with MEG data and I was wondering about the best way to combine projectors and tSSS. I know that projectors are applied when creating epochs from continuous data, but if I run tSSS before that, the projectors have already been removed (for which there is a reason Iām guessing?). At what preprocessing step is it ideal to apply projectors? Would you recommend using the āextended_projā argument of āmne.preprocessing.maxwell_filterā for this purpose?
If the projectors in question are empty room projectors, then yes you can incorporate them into the tSSS computation as part of the (discarded) external basis. You do this by passing the list of empty-room projectors as the extended_proj argument of maxwell_filter().
If theyāre some other kind of SSP projector (heartbeat, blink), then SSP projectors computed before (t)SSS was run will be no longer valid after (t)SSS has been run. This is because SSP projectors are basically like per-channel weights used to combine (noisy) channels into (less noisy) virtual channels, and tSSS changes the signals in a way that doesnāt preserve the meaningfulness/accuracy of those weights.