Functional connectivity

AFNI Message Board
1

Hi all,

I was using 3dfim+ to build the seed-based functional connectivity, but I
just want it to be focused on the cortical voxels, so what should I do? And I
also want a .txt file of the correlation matrix for the following analyses,
how can I get it? Besides, how to set the optimal value of polort?

Thank you,

Wan

2

I was using 3dfim+ to build the seed-based functional connectivity, but I
just want it to be focused on the cortical voxels, so what should I do?

You can create a mask by segmenting the brain and use the mask in your seed-based analysis. If you prefer, you can project the data on surface, and then analyze the data on surface in SUMA.

And I also want a .txt file of the correlation matrix for the following analyses,
how can I get it?

The seed-based analysis shows the correlation between the seed and every voxel in the rest of brain. What correlation matrix are you referring to?

Besides, how to set the optimal value of polort?

The “optimal” order for the polynomial fitting would have to be found through a few model building processes. The lazy (and rule of thumb) approach is to simply adopt one based on the length of each run (e.g., using “-polort A” in 3dDeconvolve)

3

Hi Gang,

Thank you, but I think I didn’t express my questions clear.

I want to form the connectivities between the hippocampus and all the cortical voxels. I’ve got the average signal from hippocampus by using 3dmaskave. 3dfim+ seems to calculate the correlation between average hippocampus signal and all the other voxels, including subcortical structures, cerebellum, white matter and CSF. How do I avoid this? Using a cortical atlas to mask the cortex out or something else?

And yes, I just want to get the correlation between the seed and every cortical voxel in a .txt file, instead of a dataset.

Thank you,

Wan

4

Using a cortical atlas to mask the cortex out or something else?

Yes, that would work.

I just want to get the correlation between the seed and every cortical voxel in a .txt file, instead of a dataset.

Try 3dmaskdump.

5

Thank you!!!