extract mean value of cluster

AFNI Message Board
1

How to extract the average value of the statistical parameter map ( For example: VMHC map) of 1000 subjects through a mask file (i.e., a cluster obtained by group analysis)?

2

Hi-

Can we clarify this first? Do you want an average of an ROI per subject, and then average of that across all subjects? Or do you want a voxelwise average acroos all subjects?

–pt

3

I only want an average of an ROI per subject.

I have obtained:

  1. a table (.csv) .
    look something like:
    Subj Group InputFile

subj0046 patient subj.0046.nii

subj0047 patient subj.0047.nii

subj0048 patient subj.0048.nii

subj0049 control subj.0049.nii
…
subj1000 control subj.1000.nii

  1. A mask file.

3.1000 VMHC map (.nii) (i.e., A R-fMRI indices, corresponds to the functional connectivity between any pair of symmetric inter-hemispheric voxels)

I only want an average VMHC of an ROI per subject.

The output table will look something like:

Subj Group InputFile mean_VMHC (ROI)

subj0046 patient subj.0046.nii xxx

subj0047 patient subj.0047.nii xxx

subj0048 patient subj.0048.nii xxx

subj0049 control subj.0049.nii xxx
…
subj1000 control subj.1000.nii xxx

4

The main command you want to calculate the average value of an INPUT_DSET within a MASK_DSET is:


3dmaskave -mask MASK_DSET INPUT_DSET

You are asking for larger scripting advice. I will take your example things literally (spacing, names, etc.)—even though you mentioned a CSV, your text file did not contain commas to separate values in a row, so neither did my example. I made this text file:


Subj Group InputFile
subj0046 patient subj.0046.nii
subj0047 patient subj.0047.nii
subj0048 patient subj.0048.nii
subj0049 control subj.0049.nii

I also made example/test 3D volumes with those names:


3dcalc -a jRandomDataset:5,5,5,1 -expr 'a' -prefix subj.0046.nii

etc. I also made a mask file to match:


3dcalc -a jRandomDataset:5,5,5,1 -expr '1' -prefix subj.MASK.nii

NB: These are not real datasets you would want to use (they are tiny 3D volumes with random numbers and a “full” mask), I just wanted some placeholders to be able to run a test script (below).

I then made+ran the following script:


#!/bin/tcsh

# just names of things
set file_old  = "file.txt"         # exists
set file_new  = "file_NEW.txt"     # new one to be created

set dset_mask = subj.MASK.nii

set nlines = `cat ${file_old} | wc -l`

# we assume that line 1 is a "header", so just copy it:
set line = `sed -n 1p "${file_old}"`
echo "${line} AVE" > "${file_new}"

foreach i ( `seq 2 1 ${nlines}` )

    set line = `sed -n ${i}p "${file_old}"`

    set dset = ${line[3]}

    set ave  = `3dmaskave -quiet -mask ${dset_mask} ${dset}`

    echo "${line} ${ave}" >> "${file_new}"
end

… by which this was created:


Subj Group InputFile AVE
subj0046 patient subj.0046.nii 0.126604
subj0047 patient subj.0047.nii -0.0212304
subj0048 patient subj.0048.nii 0.0762452
subj0049 control subj.0049.nii 0.0418862

Is that what you want? (This all depends on your mask being accurate, etc.)

–pt

5

If you have multiple mask ROIs, then consider 3dROIstats:

3dROIstats -mask mask+orig. ‘func_slim+orig[1]’