How to calculate theta power for EEG data and run statistics

Discussions
,
1

Hello, I am trying to calculate theta power for 14 subjects data for 2 conditions and want to see if the 2 conditions are significantly different. The data is already preprocessed and baseline corrected. It’s in .mat format, and structured as channels x time x trials. I can calculate the power, but how can I do a paired t-test?

2

This does not look like a file produced by Brainstorm.
Are you doing all the pre-processing in Brainstorm?

This is described in the Statistics tutorial:

However, the inputs need to be be Brainstorm PSD files (one for each trial) that contain the power for frequency bands (only theta in your case).

image
image673×221 7.2 KB

Also note that a more standard analysis is to test the magnitude (ie. sqrt(power))

3

The data was preprocessed in matlab, I believe, done by a previous lab member. I didn’t get the raw data, I just got the epochs for different conditions. So, I suppose from what you’re suggesting is that I'd need to convert the individual trials to Brainstorm PSD files.

4

That is correct. An easy way to do so would be:

  1. Concatenate all the trials in a variable X, and save that X in a .mat file
  2. Review the concatenated file as raw file, using the format EEG: Matlab matrix
  3. Import the epochs/trials by splitting using the original duration of the trial

Then you would compute the PSD (in bands) for your trials, and compute the statistical test

5

Thank you, I’ll try that approach.