Hi Alfonso and everybody,

I have a couple of question regarding the QA variables that the last versions of conn create after the preprocessing and the denoising step.

1) Which are the second level QA variables that is most relevant to include as nuisance variables in my second level analyses ? I am dealing with a sample of Parkinson patients and I see quite a bit of movement [first and third quartile of mean movement = .11 and .24; first and third quartile of max movement = .50, 1.6]. Should I routinely include all of them (I guess the answer is no, as for example invalid and valid scans are perfectly collinear, but you see my point).

2) what is the QA timeseries effect that I can use as source in the denoising step or in the first level analysis ? How is it calculated and what should it "remove" from the time-series.

As a bonus question: in the light of the fact that I have a lot of motion, would it be a good idea to regress out the global signal ? And to this aim, should I just include the grey matter mask as source in the denoising step ?

Thank in advance for any help you can provide

Alain