Hi all,

I am having trouble identifying the cause of strange FC distributions in 27 out of a 205 rest runs in my dataset (at most 2 problematic runs for any single subject; 41 subjects). These runs have a peak at zero disrupting an otherwise typical bell curve.

Confounds are specified according to Ciric et al 2017 36p+scrub protocol. Data were preprocessed in fmriprep however ICA AROMA was not used nor were any compcor regressors included (again, using 36p+scrub). This is especially confusing because the same data were previously preprocessed in a conventional SPM batch and using the same 36p+scrub protocol in conn, but nothing unusual appeared in the FC distributions. The only relevant differences between the two pipelines are that in the fmriprep version distortion correction and boundary-based registration were used, whereas in the SPM version distortion correction was not used and registration was done with conventional SPM routines. The problematic runs do not have excessive flagged timepoints nor fewer degrees of freedom than other runs. 

I am attaching the QC plots from each pipeline. Thank you for any comments and insight.

Thank you,
Walter