Hello,
I am processing an SEEG signal and I want to make sure that I am doing the correct steps in the right order.
To briefly explain the experiment - participants are presented with 4 different types of facial expression resulting in total of 120 trials (30 for each category). I want to look at (1) how the neural response to different types of expressions differs in different regions (amyg, HC, temp pole) and (2) look at Granger causality between these regions.
Here is my workflow with some questions I have.
Import anatomy -> link the raw recordings -> import electrodes coordinates -> extract the electrodes’ location -> create clusters based on the location and ROI -> import events -> inspection of the data (removing bad channels and artifacts) -> apply notch filter to the SEEG recordings -> import in database -> baseline normalisation (event-related perturbation) -> average each condition -> extract cluster values -> statistics -> connectivity
At what point should I apply the Morlet wavelet and reference electrodes?
Also, I am interested in responses at specific frequencies. Can I specify this later during calculating statistics?
Thank you for your help!
Regards,
Daniel
baseline normalisation (event-related perturbation)
I have never used this type of normalization on EEG/SEEG recordings.
This measure is in the interface for normalizing time-frequency maps.
It doesn't mean that it is wrong, I just cannot say whether this is a good or a bad idea.
If you are trying to reproduce the processing pipeline published in a specific article, make sure that this is really matching what was done in the article, and contact the authors for additional details in case of doubts.
If you are sure you want to use this type of baseline normalization, it would be better to apply it after averaging, as otherwise it would introduce strong biases between trials.
connectivity
The typical processing pipelines I know about compute connectivity measures from continuous recordings or individual trials, not averaged data.
But it might make sense to compute Granger Causality on averaged data, I really don't know. Try to stay as close as possible to the workflows used in your reference literature.
At what point should I apply the Morlet wavelet and reference electrodes?
Morlet wavelets: on individual trials, maybe followed by the ERS/ERD normalization.
Re-referencing:
Also, I am interested in responses at specific frequencies. Can I specify this later during calculating statistics?
You can't specify your frequency bands of interest on the fly during your statistical tests. First you compute the set of values you want to test.
If you are interested in one single frequency band in all your analysis, you can apply a band-pass filter on the continuous files before importing the epochs.
Or you can compute the time-frequency maps (using Morlet wavelets, or the envelope of the Hilbert transform on various frequency bands) and then run statistical tests on theses TF results.