Some Questions about AAL2

Hello everyone,
I am new to this amazing software. I have used the freesurfer’s parcellation command and it worked OK.
Now, I want to do the same but using the AAL2, but I have some problems when I use the AAL2: my connectome.csv is empty!
Here are some commands : -d 3 -o ants -f brain.nii.gz -m aal2.nii.gz

antsApplyTransforms -d 3 -i aal2.nii.gz -r brain.nii.gz -n NearestNeighbor  -t ants1Warp.nii.gz  -t ants0GenericAffine.mat  -o native.nii.gz

# labelconvert
labelconvert native.nii.gz /usr/local/matlab/R2019b/toolbox/spm12/toolbox/aal/ROI_MNI_V5.txt   \ 
	     /home/limpid/miniconda3/pkgs/mrtrix3-3.0.2-h6bb024c_0/share/mrtrix3/labelconvert/aal2.txt  \ 

tck2connectome -symmetric -zero_diagonal -scale_invnodevol fibs_200k_angle45_maxlen200_act.tck 
  native_relabel.nii.gz connectome.csv    -out_assignment assignments.csv -force

mrview brain.nii.gz -connectome.init native_relabel.nii.gz -connectome.load connectome_a2009s.csv




connectome.csv is empty:

I want to know which step is wrong, and what should I do?

Hi Sarah, -d 3 -o ants -f brain.nii.gz -m aal2.nii.gz

This looks like you are performing image registration using the parcellation data. That image does not contain anatomical information that can be used to drive image registration. Please see relevant Wiki pages.


Showing two images separately like this can be deceiving if trying to assess whether or not they overlap spatially, as the camera is not guaranteed to remain fixed when you switch from one image to the next. The Overlay tool in mrview is preferable.

connectome.csv is empty:

The fact that image native_relabel.nii.gz appears to contain only zeroes is likely the primary observation of interest; failing to produce a connectome from such is kind of to be expected.

I can’t help noticing that file aal2.nii.gz and file ROI_MNI_V5.txt are coming from two different locations on your file system. If the content of those two files do not correspond to one another, then the labelconvert process will not have the intended effect. Indeed just looking at the parcellation image in isolation, there’s a good chance that the values within that image already increment from 1 with no gaps, in which case that command is not even necessary. If the values in file ROI_MNI_V5.txt are e.g. in the thousands, then the problem is that you are using a LUT that is inappropriate for the parcellation image you are using. This page explains what this command does and why, so will hopefully also show why it could go wrong.