Hi AFNI Experts,

I have a general question that I've always wondered about. At a group level, I believe that average percent signal change extracted from an fMRI analysis would be reasonable if < 1%. Usually, I actually see values on average < 0.5% or even lower.

What about on an individual level? I'm looking at my data after extracting percent signal change values using 3dROIStats, and I'm seeing some values that look quite bizarre (i.e. > 2%) in some clusters. For a couple of subjects this is consistent across many different brain regions, so this could be motion artifact, but this isn't the case for all subjects. Sometimes there are values > 1 % in just one of many regions. Is there an absolute value cutoff that should be used to exclude participants from analyses? Is this appropriate to do post-hoc? There are no indications (i.e. motion criteria we currently use or anything else) that would have suggested these participants should have been excluded from the analyses to begin with.

I'd love to hear thoughts on what is appropriate and if people use cut-off values when they inspect their data.

Thanks!
Anita