ERD/ERS and baseline normalization

Discussions
1

Hi Francois,

I have a question that is maybe a little more complicated.
I have several conditions, and one of condition is used as a “control”. I would like to do time frequency analysis comparing various conditions, and also several conditions compared to the “control”.
I think that for comparing the “control” and other conditions, it would be appropriate to use ERD/ERS, but I am not sure if I would do ERD/ERS on baseline normalized data. Would it be ok?
This 'control" could be considered as a baseline.
I hope the question makes a sense. Basically, I am not sure if it is appropriate to do ERD/ERS on baseline normalized data.

Thank you again.
Svjetlana

2

Hi Svjetlana,

The cases of uses of the process “ERD/ERS” process are exactly the same as the “Z-score” process. They just give a different type of standardization (google them for examples of application cases).
In the case of an event-related study, you can use those two processes to standardize the post-stim values based on the pre-stim values. The process is the following:

  • Import your epochs (including a pre-stim preriod as long as possible)
  • Optionally calculate the sources (if you work in source domain)
  • Calculate the average of the time-frequency maps for each trial (Morlet wavelets or Hiler
    => you can do this in one process call only, check the time-frequency tutorial [Process1 tab]
  • Standardize the TF values: post-stim vs. pre-stim (Z-score or ERD/ERS) [Process1 tab]

To compare time-by-time the results between a standard and deviant condition (or a event of interest vs a control condition), you should rather use the menus in the Process2 tab.
If you have 100 events of interest and 50 control trials:

  • Import all your epochs (100+50) with the same time definition in the database
  • Optionally calculate the sources (if you work in source domain)
  • Calculate the time-frequency decomposition for each 100+50 trials (Morlet wavelets or Hilbert transform)
  • In the Process2 tab, drag the 100 trials in FilesA and the 50 controls in FilesB, select the button “Process time-frequency” and run a t-test

Cheers,
Francois

3

For more information on the ERD/ERS process, see the older posts of this thread:
http://neuroimage.usc.edu/forums/showthread.php?1047

4

Hi Francois,
I have some other questions about this.
It seems that for comparison, we need all trials.
I would like to do between group comparison on my data, so i am not sure if it is possible to do it in Brainstorm.
Originally i was thinking to do ANOVA with z-scores, comparing groups and conditions.
I am not sure if you can say how it would be best to do it in Brainstorm, using t-test, or exporting z-scores into SPSS maybe and do ANOVA?
Thank you again
Svjetlana

5

Hi Svjetlana,

It depends on the test you want to run. If a t-test is enough for you, then it’s much easier to stay within the Brainstorm interface.
If you want to run more complex tests, then export your source/scouts values and process them with another software (SPSS, R, etc)

Cheers,
Francois

6

Thank you Francois.