3dNetCorr On Subset of the Time Series

AFNI Message Board
1

Hello AFNI people,
I am using 3dNetCorr to compute correlation matrices on time series covering 13 minutes MRI runs, each with 780 EPIs (TR = 1sec).
I would like to compute the correlation matrices over 10 minutes only instead of the full 13 minutes, which basically means telling 3dNetCorr to use only the EPIs 1 to 600 instead of the full series. I looked into the doc and the forums but couldn’t see anything like it.
Is there anyway to do this easily in AFNI without having to save a bunch of new files with only the EPIs of interest?
Thanks in advance!
Freya

2

Hi, Freya-

Sure, we have the technology!

If this is an example of running 3dNetCorr on the full time series (some made up names here, to protect the innocent dsets):


3dNetCorr                                       \
    -inset FMRI_DSET.nii.gz                 \
    -in_rois ROI_MAP.nii.gz                  \
    -fish_z                                     \
    -prefix o.full                   \

… then here is an example of selecting “just” the first 600 time points from the input FMRI dset:


3dNetCorr                                       \
    -inset FMRI_DSET.nii.gz'[0..599]'                   \
    -in_rois ROI_MAP.nii.gz                  \
    -fish_z                                     \
    -prefix o.part                   \

This is using the “subbrick selector” functionality in AFNI. If you run this command to open the given afni_handout:


afni_open -aw afni30_vol_dset_basics.pdf

… then go to PDF page 24 for more examples. Note: Some kind of quotes around the square brackets is necessary, because otherwise the Linux shell will try to interpret those as its own special characters.

Note also that AFNI uses zerobased counting, so that the “first” volume in a dset is actually volume [0]. This is in line with programming languages like C and Python. (Soooorrrry Matlab and R users :(…)

–AFNI person “pt”