Closed adam2392 closed 2 years ago
All of the spectral estimators IIRC currently collapse across epochs to make their estimates (i.e., there is no estimate of connectivity for an individual epoch), so I don't see a way to do this.
(and currently I don't think they'd return SpectroTemporalConnectivity
but rather just SpectralConnectivity
even now, since the temporal dimension is converted / changed to spectral domain?)
This is now closed by #69 and #67 introducing spectral connectivity over time.
Inside mne_connectivity.spectral_connectivity, all functions assume that Epochs are samples of the same connectivity structure. That is, they average over Epochs implicitly in the function.
The code is a bit harder to parse through, but wondering if it is worth the effort to also, allow it to return an
EpochSpectroTemporalConnectivity
, rather thenSpectroTemporalConnectivity
to allow the user some flexibility of later "combining" over epochs with median/mean/etc?