Connectivity matrix ABCD

Hi all! I have a general question on functional connectivity analysis from rsfMRI. I find the correlations between networks and ROIs but I was looking for the correlation between single ROI (for instance connectivity matrices from all Desikan ROIs): is it available? I need to construct the correlational matrix. Also time series would work! Thanks


I’m a little confused by the question. I don’t see how you could calculate correlation “between a single ROI”: correlation inherently using 2 time series to be calculated.

Just to check: could you be a bit more specific about what it means that you have found “correlations between networks and ROIs”?

You mention the Desikan atlas—does that mean you have run FreeSurfer, and have all the ROIs on the same grid as your rsFMRI data?

You could use AFNI’s 3dNetCorr to calculate the correlation matrix of those time series and ROIs (and I added the “-fish_z” option in case you want both the Pearson r matrix, as well as FIsher-Z converted matrix):

  3dNetCorr                          \
      -inset DSET_FMRI               \
      -in_rois DSET_ROIS             \
      -fish_z                        \
      -prefix OUTPUT

That would accomplish calculating the correlation between each possible pair of ROIs.

You could also add -ts_wb_corr to get the whole brain correlation maps of each ROI’s average time series with the whole brain.

If someone asked me about calculating “correlations between networks”, I might think they wanted to calculate spatial correlations between two volumetric maps, such as with 3dTcorrelate.


Dear expert,

thank you for your kind response. My question was related to the fact that the ABCD study makes available the correlational values between the Gordon Networks and the FreeSurfer subcortical parcels, as well as the temporal variance values of all the ROIs in the Desikan and Destrieux atlas. In this regard, I was wondering if there are also available either the timeseries of such ROIs or their correlational values.

Ah, I understand now—thanks for describing that. I am not familiar with the specific processed information they provide, and instead was describing how to calculate things from the raw-er data itself.

Well, if the time series are available then perhaps some of my reply will could be useful.