[Eeglablist] Frontal Alpha Asymmetry calculation & Preprocessing

Krogmeier, Claudia M ckrogmei at purdue.edu
Thu Mar 18 13:10:57 PDT 2021


Hi list,

I am interested in analyzing frontal alpha asymmetry from my continuous EEG data. In my study, I want to understand the FAA score during stimulus presentation of different events. I present events for 6 seconds, with 5 seconds rest in between. Below are my steps:


  1.  compute average reference
  2.  high pass filter at 2Hz
  3.  low pass filter at 30 Hz
  4.  use ASR plugin to clean data
  5.  compute average reference
  6.  interpolate channels lost if any
  7.  split data into epochs based on event markers (-1 to 6 second epochs)
  8.  epoch rejection based on voltage threshold of +- 75 uV
  9.  interpolate channels lost if any
  10. Compute FAA
     *   get power spectrum from F7, and F8 (8-13 hz), using ERSP: newtimf function in command line
     *   ln(F8 alpha power) - (F7 alpha power) to get power spectrum at that electrode pair

I have several questions related to preprocessing, and some questions specific to FAA.

Should I use ICA instead of ASR? (I have seen both methods in other FAA work). I have a lower number of electrode channels, so I was not sure if using ICA would be appropriate.

I have read some papers suggesting interpolation is not good for EEG data, but it seems it may be necessary to get an unbiased FAA score.  Is interpolation necessary for FAA calculations? Should interpolation be done whenever any electrode channel is deleted from the data (due to artifacts)?

Would ERSP calculation be better than spectopo for this study's case? Would ERSP output the power spectrum I need from my channel of interest?

Lastly, when I do this processing, my asymmetry values appear to be between .9 and 1.9. In the literature, I have often seen FAA values ranging from -.5 to .5.  Should my asymmetry values be negative? What would be an expected FAA range? Currently, I am analyzing a single subject, to ensure I have expected values and preprocessing steps correct before I calculate a study with multiple subjects.

Thank you so much for any references, responses, and feedback. I really appreciate it!
Claudia




More information about the eeglablist mailing list