Compare source between hemisphere

[xiongruochu]

hello，

I'm considering comparing the source differences between left and right hemisphere within same experimental condition and same time window, using paired t-test. (for example, brain activation of left insular VS activation of right insular)

I referred to official tutorials, but that was comparing between different experimental conditions. (different with my task design; my design only has 1 condition)

And I also referred to this topic (Comparing two hemispheres) , but I don't understand the steps.

Is there anyone who can help me with the analyzing steps?

[Raymundo.Cassani]

You can do that by using Scouts:

1. Create scouts for the regions that you want to study: RegionA and RegionB
   https://neuroimage.usc.edu/brainstorm/Tutorials/Scouts
   

2. On the source files for each trial, use the process Extract > Scout time series to extract the time series for the RegionA scout
   

   

3. Repeat step 2 for RegionB

4. At this point you should have a 2 matrix files for each trial, one with the time series for scout RegionA and one for the scout RegionB. And you can run the paired t-test as indicated in the Statistics tutorial: https://neuroimage.usc.edu/brainstorm/Tutorials/Statistics

   

[xiongruochu]

Thank you for the answer. Sorry to bother you again, but I have two more questions.

1,

@Raymundo.Cassani

use the process Extract > Scout time series to extract the time series for the RegionA scout

Can I extract several scouts at one time?
Or I'm forced to extract 1 scout at each time?

2,

@Raymundo.Cassani

At this point you should have a 2 matrix files for each trial, one with the time series for scout RegionA and one for the scout RegionB

I have 22 subjects and I want to do the statistical test using all subjects data (scouts for each subjects are the same).
After I extract all the scouts I want, should I need to do average within subject (get the average scout source for each subject) first, and then normalization, and projection onto template or anything?
Or I just drag all those extracted scout source of all subjects into process tab and run paired t-test? And then project the statistical results onto template?

Looking forward to your response.

[Raymundo.Cassani]

@xiongruochu

Can I extract several scouts at one time?

Yes, you can extract for multiple scouts at the same time. Time series for the scouts RegionA and RegionB will be in the same file.

@xiongruochu

Or I'm forced to extract 1 scout at each time?

However, if you want to compute statistics of for time series of RegionA vs RegionB. You must have a those time series in different files. Thus, yes it is needed to extract 1 scout at the time.

@xiongruochu

After I extract all the scouts I want, should I need to do average within subject (get the average scout source for each subject) first, and then normalization, and projection onto template or anything?

You could follow the same recommended workflow as when working with sources: this is to say,

1. First compute source average at Subject level
2. Extract the Scout series
3. Normalize the scout time series. At this point different subjects will have the source activity in the same range of values

Projection of the sources (after step 1) is not necessary if the Scouts come from a Atlas. In that case, the Scouts are already a projection from a Default anatomy (where the Atlas is defined) into the Subject anatomy, so the scouts represent the same regions across subjects.

[xiongruochu]

@Raymundo.Cassani

• First compute source average at Subject level
• Extract the Scout series
• Normalize the scout time series. At this point different subjects will have the source activity in the same range of values

If I want to extract the activity at a specific frequency, can I extract a small time window (very narrow band-pass filter) of the scout series after step 2? Or do the filter after normalization (to avoid “an underestimation of the baseline variance“ according to the Workflows tutorial) ?

[xiongruochu]

Sorry to disturb again.

I ran the parametric paired test for one scout (L vs R), and then the warning appeared:

What could be the reason for this?