PSD comparison (continuous EEG data)

Discussions
1

Hello Brainstorm team!
We are working on continuous EEG data for a within subject group analysis. We are interested in a variety of frequency bands through all the cortex. We want to compare the two conditions in how they differ from a baseline.
So, for each participant, we have: baseline recording + experimental condition A + experimental condition B. We produced 3 PSDs by participant (for the baseline + A + B for), and Z scored the PSDs of A and B with the baseline.
Then, we computed a paired t-test analysis (FDR correction) using the normalized A and B PSDs.
What I would like to ask is, first, does this seem to be a proper way to deal with this?
Secondly, we would like to verify the adequate way to interpret the directionality. Because we did not go in the source domain, are the t-values to be interpreted as is or do we have to add another step?

Thank you so much for your help and have a wonderful day!

Alexandra

2

**We originally wanted to project in the source domain, but it asks for 12 GB RAM and create Z scored source file of 10 GB each. I guess it is because the sources files are computed on the whole 7 minutes of data instead of an average, but an average would not be informative in our case. Can you think of any other way to Z score PSD in the source space?
Would Z scoring PSDs that were computed in source space work?

Thanks!

3

Hi Alexandra,

What I would like to ask is, first, does this seem to be a proper way to deal with this?

It looks OK to me.
@Sylvain @pantazis : Can you confirm?

Because we did not go in the source domain, are the t-values to be interpreted as is or do we have to add another step?

When interpreting t-values on strictly positive values, positive values should be interpreted as A > B (A=list of files on the left, B=list on the right)

Can you think of any other way to Z score PSD in the source space?

You could refer to our resting state tutorials:
http://neuroimage.usc.edu/brainstorm/Tutorials/RestingOmega
http://neuroimage.usc.edu/brainstorm/Tutorials/HCP-MEG

4

Looks good to me.
I am nor sure what you mean by “directionality” though.