Combing ROIs: error with 3dcalc

AFNI Message Board
1

Hi, List experts,

I had problems during the ROI combination, but I do not know how to understand the error alarms below. Is that I have problems with the ROIs themselves or anything else?

Thanks so much,
Juan

[S01@NY_Net] $ 3dcalc -a MPFC.nii -b VMPFC.nii -expr ‘(a+b)’ -prefix PFC.nii
++ 3dcalc: AFNI version=AFNI_16.3.13 (Nov 23 2016) [64-bit]
++ Authored by: A cast of thousands
** FATAL ERROR: dataset VMPFC.nii differs in size [153594 voxels] from others [902629]
** Program compile date = Nov 23 2016

[S01@NY_Net] $ 3dcalc -a L_VS.nii -b R_VS.nii -c rMPFC.nii -d rVMPFC.nii -expr ‘(a+b+c+d)’ -prefix 4ROI_Merged.nii
++ 3dcalc: AFNI version=AFNI_16.3.13 (Nov 23 2016) [64-bit]
++ Authored by: A cast of thousands
** ERROR (nifti_image_read): failed to find header file for ‘rMPFC.nii’
** FATAL ERROR: can’t open dataset rMPFC.nii
** Program compile date = Nov 23 2016

[S01@NY_Net] $ 3dcalc -a MPFC.nii -b VMPFC.nii -expr ‘(a+b)’ -prefix PFC.nii
++ 3dcalc: AFNI version=AFNI_16.3.13 (Nov 23 2016) [64-bit]
++ Authored by: A cast of thousands
** FATAL ERROR: dataset VMPFC.nii differs in size [153594 voxels] from others [902629]
** Program compile date = Nov 23 2016

2

It seems that I need first resize and make them same in the size…

3

That’s right. Glad you figured it out. Just in case, you may want to use this atlas for the Ventromedial prefrontal cortex, described here:

https://afni.nimh.nih.gov/afni/community/board/read.php?1,147025

4

Thanks so much, Daniel.But do you have any suggestion to use that VMPFC (In that post, you introduce the paper and we can download the mask, but it includes a lot of small masks of VMPFC). They are MNC formatted files, but AFNI can read it. Does it mean that if I want to only use the general VMPFC, I need first combine all of them? Thanks so much!!!

5

That atlas should be distributed by default with AFNI as the dataset, MNI_VmPFC+tlrc (.HEAD and .BRIK). Yes, you could combine all sub-regions together. depending on what you want to do. In 3dcalc, that would just be a single step operator in the expression for any non-zero, positive value in the atlas.

Also looking back at your 3dcalc expressions, you may want to consider either a ‘step(a+b+c+d)’ or ‘step(a)+2step(b)+4step(c)+8*step(d)’ as the expression. The first gives either a 0 or 1 at each voxel depending on whether any of the regions overlap that voxel. The second expression gives a binary-like code depending on which regions contribute to that voxel, and each resulting value comes from a unique combination of overlapping regions.