Dear list,
recently, I have been thinking about how to reduce movement artifacts in my first level standard GLM analyses in fMRI as well as in psychophysiological interaction (PPI) analyses. I came across global signal correction, which according to some sources (e.g. Power et al. 2014) appears to reduce movement-related artefacts in the data. Additionally, the volterra-extended realignment parameters (i.e. the first derivatives of the six realignment regressors, as well as the squares of the original regressors and the squares of their first derivative) are used in some fMRI analyses.
I searched for mailing list threads or methodological papers comparing different approaches to correct motion-related signal changes. However, those articles I found (such as the one by Power et al, 2014), are related to the effects of motion correction on resting state functional connectivity (not PPI).
So my question is: which (if any) considerable benefits or drawbacks are there to using Volterra extensions of realignment regressors and / or global signal correction when conducting PPI analyses?
Furthermore: in which way is global signal correction (by introducing the global signal as a regressor in the design matrix) comparable to the 'scaling' option one can choose under 'global normalization' when setting up the first level analysis?
I'll be thankful for any helpful answers and literature suggestions!
Best,
David
|