How to explore correlations between source activations and behavior measures

Discussions
1

Dear Brainstormers,

I wanted to do a brainstorm version of this pipeline that I did in BESA:

image
image1420×601 140 KB

https://www.nature.com/articles/s41598-019-47643-1/figures/2

correlation of each voxel to a certain external measure across subjects.

I can do this with scout activations (export scout time series and then run correlations), but I wanted to "explore" the data without any ROIs and just look at the peak correlations. Any suggestions? I suppose I could use an Atlas that covers the entire brain, and run the correlations on those ..

sorry if this has been covered before .. I couldn't find on the forum

thanks
andrew

2

With the processes Correlation 1xN (Process1) or AxB (Process2), you can compute correlations of all the source time series with an external signal, and obtain cortical or volumes maps of correlation values.

1 Like
3

Hi Francois,

Just following up on this. How do you suggest I get the "external signal" into Brainstorm? It is not embedded in the subject folder. In my case, the external signal would be a single number for each subject (e.g., performance on a task). I've done this before in FieldTrip where I have the .pow for each subject and a single "external value" for each subject, and just run the correlation this way.

thanks
andrew

4

The correlation processes in Brainstorm compute correlation coefficients between two signals. This requires to have multiple time samples, and the same number of time samples in the two signals you want to compare.

For any other measure, you might need to do this from custom Matlab scripts.
This tutorial can help you to get started with scripting with Brainstorm:
https://neuroimage.usc.edu/brainstorm/Tutorials/Scripting