It is possible to average topography for continuous EEG (about 1 minute registrations)?

Discussions
1

Hi guys,

I'm using 10-20 pattern, collecting 19 channels and i need to average lots of files of continuum EEG for different frequency bands. I've found the procedure for ERPs in the tutorial... I do not know if i can use the same procedure for continuum EEG.

Thanks in advance.
Marcos Rosa
UFMG/Brasil

2

You can technically do all the averaging you want on EEG files in Brainstorm, across time, across electrodes, across files. Not everything is meaningful, though...

  • Averaging over time long EEG continuous recordings does not make sense: the EEG potentials fluctuate around zero, and the sum in time would tend towards zero.
  • Averaging time-frequency decomposition of a single file over time would not be zero, the time average would be meaningful, but this is a very convoluted way to compute a power in a given frequency band. Instead, use the PSD process for that.
  • Averaging continuous EEG across multiple does not make sense, as it would average effect patterns of brain activity that are not aligned in phase.

What are you trying to achieve?

3

Thanks for your response. It helped me as I am not an EEG expert.

I'm trying to plot an average topography of continuous EEG files. Some 60 second files.
I'm trying to show the average brain activation in these files, focusing on different frequency bands.
So I think I can get some good results using PSD on the files and then plotting the 2D disk.
Let's suppose I do this for the Alpha band. It'll give me the topography I need for the Alpha band, right? In this case, what is best to measure? Power; Magnitude; or Log(power)?

Thank you again!
Marcos Rosa

4

Let's suppose I do this for the Alpha band. It'll give me the topography I need for the Alpha band, right?

In the advanced options of the PSD, you can select how you want the frequency bins of the FFT to be averaged in larger frequency bands.
https://neuroimage.usc.edu/brainstorm/Tutorials/ArtifactsFilter#Evaluation_of_the_noise_level

In this case, what is best to measure? Power; Magnitude; or Log(power)?

The power is probably the most used measure.
But you should probably start by trying to reproduce what is done in your reference literature.

5

Thank you for your support.
Your information and tips truly helped me.

Regards,
Marcos