smallest seed size fmri

AFNI Message Board
1

HI everyone

I would really appreciate if someone can weight on this: what is the minimum reasonable seed size for resting state fmri (seed to all brain) timeseries extraction. (I’m using HCP ICA cleaned data )
I am trying to correlate tractography findings with resting state fMRI connectivity (seed to all brain) from the thalamus . I have basically parcellated the thalamus using tractography and now I want to see if a similar antero-posterior pattern is maintained also in the cortical networks at rest . I have small masks of about 100 voxels each but when I use 3dmaskave and 3dfim+ the resulting correlation bucket requires a very low threshold of the Afni slider (around .2) to see some activation. Is that expected?

thanks a lot

2

I have small masks of about 100 voxels each

What is the voxel size? 100 voxels does not sound like a small seed size to me.

but when I use 3dmaskave and 3dfim+ the resulting correlation bucket requires a very low
threshold of the Afni slider (around .2) to see some activation. Is that expected?

Are you referring to the correlation value between the seed and the rest of the brain? If so, that is not necessarily small. Try thresholding with the t-statistic that is associated with the correlation value.

3

Thank you Dr Gang for your answer.The thresholding was indeed the problem.

Best regards