Hi,

I have been using the scripts to analyze my own data as well as some of the FCON data examing the single subject RSFC. Everything seems to work OK but I had expected a stronger Z value between (for example) the PCC and anterior medial frontal areas etc whereas I get values of Z = 1 or so.  The maximum Z value at the subject level does not seem to go over Z = 2. The spatial pattern in the RFSC matches what I had expected.

I haven't modified the scripts - well almost - I am using the AFNI commands to remove the variance in the resting state time series due to the 6 motion corrections regressors, the global, WM and GM signals.
(i.e. 3dREMLfit  -input ..... -matrix nuissance.xmat.1D -mask .... -Rerrts res4d.nii.gz -GOFORIT ). I used this modification from a previous post on this forum.

In the above command - I usually get a warning about having 2-3
covariates being collinear (in the above command) and has left me
wondering if that is causing bigger problems than ... The covariates are usually the motion parameters or may include the global signal covariate.

Question: are the Z value range at the subject level that I am obtaining similar to what others have obtained?

All the best,

Arun

Hi Arun,

these Z values that you get from the RSFC scripts are not Z-stat, but Fisher's Z correlation coefficient values, so the values that you get are pretty normal. Z=1 is Pearson's r=0.762, so it is a pretty strong correlation, and Z=2 is Pearson's r=0.964, so it is almost a straight linear correlation which is statistically significant. Your values at the subject level are all right.

Cheers,
Michal

Hi Arun,

as Michal indicates the Z-values from the RSFC scripts are Fisher Z transformed correlations.

However, I would worry a bit about the message you get about some variables being a linear combination of each other during nuisance regression. That does not sound correct and might induce problems for your regression. What does your model look like that you get such warning with the nuisance signals that are usually very different from each other (i.e., not prone to being linear combinations of each other).

Best,
Maarten