Dear Support,

I am processing 1.5 T T1w fMRI resting state data, doing preprocessing in CONN with default pipeline and settings. And getting different ROI-to-ROI first level results every run. Also, these results differ by a machine (PC) there I am computing.

I've made an experiment: took 2 fMRI scans and computed the ROI-to-ROI matrixes for 164 Seed regions one by one, together and on the different computers- all results differ

Why is it so? How it can be overcome and where this randomisation occurs?

Thanks so much