[Eeglablist] When to apply pop_rmbase

[#HENG JIAMIN GLADYS#]

Dear EEGLAB list,

I have a query as to when to apply baseline correction via the function "pop_rmbase".


I am working on a dataset that is looking at extracting the power spectra in alpha/theta/beta frequency bands during music listening. Participants listened to several musical excerpts (each 45 sec) with a 15 sec break in between music stimuli.



In my preprocessing pipeline, I included a step that did a baseline correction of -1 second to each epoch (i.e. individual music clips) because "Removing a mean baseline value from each epoch is useful when baseline differences between data epochs (e.g., those arising from low-frequency drifts or artifacts) are present." (https://urldefense.proofpoint.com/v2/url?u=https-3A__eeglab.org_tutorials_07-5FExtract-5Fepochs_Extracting-5FData-5FEpochs.html-23removing-2Dbaseline-2Dvalues&d=DwIF-g&c=-35OiAkTchMrZOngvJPOeA&r=kB5f6DjXkuOQpM1bq5OFA9kKiQyNm1p6x6e36h3EglE&m=auzm18_BF9WNzeSRvFPDKEM_SiUEsNWpxiyza9uzEXs&s=M3eOWlwNSQ45fvEuz0e7s1Iz5guLQT6oflUFx_DL_4Q&e= )



As such, following is my pipeline:

  1.  Bandpass filter 1 – 30 Hz
  2.  Remove noisy channels (threshold: above kurtosis 5)
  3.  Average re-reference
  4.  Epoch: baseline correction of -1 second (i.e. -1 sec to 45 sec)
  5.  Detrend
  6.  Re-epoch from 0 sec to 45 sec
  7.  Apply ICA
  8.  Remove ICs related to eye-blinks and muscle artefacts using MARA
  9.  Remove EOG channels
  10. Interpolate
  11. Extract power spectra values for each epoch

Is it appropriate to apply the baseline correction of -1 second before epoch (i.e. Step 4)? Or should the baseline correction be done after calculating spectral power values, and not at the raw EEG values (i.e. Step 11)?

I would greatly appreciate any reference to a paper that shows the theoretical basis of why doing baseline correction before epoch is okay.

Thank you very much in advance!

Best regards,
Gladys