You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This can be something useful to denoise brain signals indeed.
In the same vein, for this very specific branch of sktime you should consider referential vs differential recordings:
Pair-wise local referencing could also multiply the dimensions (using esentially the same module as the object of the issue)
That means compute the difference between pairs of adjacent electrodes -provided that spatial information is available- either on a human cap or on multi-channel probes. This could be could be also called local referencing. It implies that you would have signals of much lower absolute amplitudes (given that they will be very similar between adjacent channels) but they will carry a much more local information, different frequency bands, more transient activity and possibly providing information of multi-unit activity at higher frequencies (action potentials from un-discriminated neurons [when recorded intra-cranially])
Looking into standard preprocessing pipelines, it would be good to add common average re-referencing.
See article here: https://eeglab.org/tutorials/ConceptsGuide/rereferencing_background.html
This could be added as a series-to-series transformer
The text was updated successfully, but these errors were encountered: