Hi Francois,
I have PSD files in three conditon (task A, task B, and rest) in 23 participants, following the tutorial "Resting state recordings from the OMEGA database". I would like to compare the PSD in task A with that in task B, taking into account the rest condition. I think that there are two option, one is anova in three condition, the other is normalize the PSD in two tasks based on that in rest, then use paired t-test. I think former is complicate in terms of considering the freq, vertex, and post-hoc test. so I think later is better solution. In process 2, there is Baseline normalization (A= baseline) option in Standardize. But I just want to normalize PSD in task A and B based on total power of PSD in rest condition, similar to Relative power (divided by total power) option in Spectral normalization, but there is no such option in process 2. How can I do such normalization? or Is the PSD normalization based on the PSD in rest condition wrong?
Option 2: Division by the total power in the rest condition
This is not possible from the interface either, you would also need to compute this from a script.
Option 3: t-test (PSD(A)-PSD(rest)) vs. (PSD(B)-PSD(rest))
I think that this is what I would try first.
Use Process2 to compute the difference of PSDs for each subject and each condition, then use process2 again to run the test between the differences.
However, I'm not sure if it would be best to compute a relative power a) before the subtraction, b) after the subtraction or c) not at all.
I recommend you go the second route and use the Normalize option (A-B/A+B) of the Difference process folder once you have selected the Condition A/B and Resting PSDs in the Process2 tab.
You can then test for differences between the normalized PSD A/B using non-parametric t-tests with permutations.
Hi Sylvain,
Thanks for your prompt reply. In case of using the Normalize option (A-B/A+B), selected PSDs are files that are computed Spectral normalization (Relative power) or un-normalized ones?