[Eeglablist] Advice on filtering strategy before ICA (EEGLAB pipeline; FAA, beta/alpha ratio/ PSD)

[Евгений Машеров]

I'd point out another reason not to use connected references (this applies to an analog connection, where the wires from the references are physically connected). We test the quality of the contact by achieving low resistance, but if the contact on one side is good (for example, 10 kOhm) and on the other excellent (1 kOhm), then we get a voltage divider with 1:10 arms, and the potential on the common wire is 90% dependent on the signal from the contact with the lower resistance. In electrocardiography, when recording signals from chest electrodes V1-V6, a Wilson terminal is used to account for this effect (connecting the electrodes of the three limbs through resistors with a sufficiently high resistance, for example, 50 kOhm), but, as far as I know, this is not used in EEG.
With digital averaging (when signals from the mastoids, ears, etc. are recorded relative to a common reference and then the arithmetic mean is calculated), this effect is eliminated.

Eugen Masherov

> Hi Kathryn,
> 
> On linked mastoids, I agree with Makoto. Based on work on alpha asymmetries, it is strongly recommended not to use linked mastoid referencing for two related but distinct reasons:
> 
> First, linked mastoids suffer from alpha “mirroring”: occipital alpha dipoles produce activity of opposite polarity at sites 180 degrees apart, and since spectral power summarizes oscillation magnitude without regard to polarity, occipital alpha bleeds into frontal leads through the reference. This contaminates frontal alpha asymmetry measures with signal unrelated to frontal sources (Smith, E. E., Reznik, S. J., Stewart, J. L., & Allen, J. J. B. (2017). Assessing and conceptualizing frontal EEG asymmetry: An updated primer on recording, processing, analyzing, and interpreting frontal alpha asymmetry).
> 
> Second, if a lateralized neural source drives the two mastoids in opposite polarity, linking them creates a reference that is itself contaminated by the signal of interest, which can suppress or distort asymmetry scores (Hagemann, D., Naumann, E., & Thayer, J. F. (2001). The quest for the EEG reference revisited: A glance from brain asymmetry research).
> 
> The CSD (current source density) transformation avoids both issues by computing the second spatial derivative of voltage, attenuating distal volume-conducted sources and providing a reference-free estimate. Although CSD also has downsides, which were discussed in a recent eeglablist thread. I can try to find it if you want to learn more.
> 
> You can perform CSD in one line in eeglab with this code: https://urldefense.com/v3/__https://github.com/amisepa/csd_transfrom__;!!Mih3wA!D-IJVG1iQfvM3mGG1a0VAvcyK6Xydhoi-GPhWN60iDa7XwfiAWK7YvhCoTrhdKrE-kTPfcJ7V2ot9FMsWN8rw9he4g$
> 
> Or with Rest/infinity method here:
> https://urldefense.com/v3/__https://github.com/amisepa/reference_infinity__;!!Mih3wA!D-IJVG1iQfvM3mGG1a0VAvcyK6Xydhoi-GPhWN60iDa7XwfiAWK7YvhCoTrhdKrE-kTPfcJ7V2ot9FMsWN8XO01aNw$
> (Also available in eeglab’s extension manager, search ref_rest_cmd)
> 
> Cedric
>