BioSemi.nl

Hi,

We've been recording EEG (64-electrode cap), GSR, Plet and Respiration.

Although concurrently recorded and saved into one .bdf file, we are analyzing the EEG data separately from the physiology data, since the recommended re-referencing during import (via the biosig toolbox in EEGLAB) seems to impact both the EEG channels and the data from the auxiliary channels, which is not ideal.

However, when only importing the physiology data (without specifying a reference in pop_biosig), 'common' is still mentioned in the reference field of the datastructure in MATLAB (as in 'common mode', and not 'grand/common average' I suppose?). Just to be sure: it is OK to ignore this feature for the GSR, Plet and Respiration channels, right? Since the (absence of a) scalp reference should play no role during the collection of these auxiliary signals.

Thanks for your help,
Jellina

You are correct: referencing should only be applied to the EEG channels, the sensor channels should remain unreferenced (or else the EEG reference is subtracted from the sensor channels).

Best regards, Coen (BioSemi)

Thanks for the previous information Coen. I do however have a follow-up question to this issue.

I know that since the BioSemi EEG signal is collected in 'common mode' (i.e. using active electrodes), it is generally recommended to re-reference first thing in the preprocessing pipeline to account for 40 dB of SNR.

However, I'm planning to analyze my EEG data in CSD (surface Laplacian) mode, which transform the data to a reference-free format and reduces volume conduction/increases spatial topographies. This transformation is most often applied at the end of the preprocessing pipeline. Is this CSD transform also sufficient to cancel out the inherent noise in the BioSemi data? Or do I still need to explicitly re-reference my data to mitigate the loss of SNR first?

Thank you in advance for your help!

Best wishes,
Jellina

Processing that only looks at the differences between electrodes (like CSD) should be unaffected by referencing. However, it would be easy to check (test whether the CSD outcome is equal with and without referencing ?).

Best regards, Coen (BioSemi)