Dear FSL experts,
I am doing resting-state functional connectivity study using seed-based connectivity methods.
For sanity check, I placed the seed in PCC region for baseline conditions (n=15) to see whether the Default Mode Network appears.
My pipeline uses first three Principal components from both WM and CSF region masks each (eroded), along with 6 movement regressors as nuisance regressors - With the seed timeseries, these regressors comprise a design matrix for the GLM analysis. This design matrix attempts to follow the compcor method.
The problem
- After running second level group analysis, the zstat map shows high zstat values over the entire Gray Matter in the brain, the values being well over the conventional threshold 2.3. It is impossible to discern DMN from fslview with the threshold of 2.3
- Only when the lower threshold is set above 4~5, I can start to see the expected DMN structure. Somehow, the zstat values seem 'biased'
What would be the reason for the high zstat value? I am well aware that '2.3' is merely a conventional threshold for zstats, but I am concerned that something is incorrect with my pipeline as it produces seemingly 'biased' zstats.
I would really appreciate any thoughts!
Thank you very much,
|