AFNI version info (afni -ver
): 24.0.08
Dear AFNI developers:
I have downloaded Schaefer-Yeo atlases from https://afni.nimh.nih.gov/pub/dist/atlases/SchaeferYeo/afni_sydist_atlas_v1.0.tgz. In the unzip directory, there are two sub-directories containing surface atlases. I check the surfsets directory and find errors in the right hemisphere files. The std.141.rh.* atlases contain the left hemisphere data. This is an example:
# transform binary data to asc
ConvertDset -o_niml_asc \
-input std.141.rh.Schaefer2018_100Parcels_17Networks_order.smooth3mm.lbl.niml.dset \
-prefix test
# part of output
<AFNI_dataset
dset_type="Node_Label"
self_idcode="AFN_lsuFpXfxA6Zj7SzbOJCC9Q"
domain_parent_idcode
geometry_parent_idcode
filename="test.niml.dset"
label="test"
ni_form="ni_group" >
<SPARSE_DATA
ni_type="int"
ni_dimen="198812"
data_type="Node_Label_data" >
29
40
10
15
42
6
45
39
14
5
23
11
29
29
29
29
29
29
29
29
29
29
29
0
0
0
0
0
...
Then I try to produce these surface files by myself. This is my script:
for hemi in "lh" "rh"; do
3dVol2Surf \
-spec std.141.MNI152_2009_${hemi}.spec \
-surf_A smoothwm \
-surf_B pial \
-sv MNI152_2009_template_SSW.nii.gz \
-grid_parent Schaefer_17N_100.nii.gz \
-map_func nzmode \
-f_steps 10 \
-f_index nodes \
-out_niml std.141.Yeo_17N_100_${hemi}.niml.dset
3drefit -copytables Schaefer_17N_100.nii.gz std.141.Yeo_17N_100_${hemi}.niml.dset
done
However, If the map_func is "nzmode", the rh_* atlas will contain values 28, which should belong to left hemisphere. After changing map_func to "mode", there will be no outliers in rh_* files. I can not determine the cause. I hope some expert can provide advice.
Finally,I have two more small problems.
First, I know MNI152_2009_template_SSW.nii.gz is based on MNI152_2009c_asym. But which template is the Schaefer-Yeo atlses provided by AFNI based on? MNI152_2009c_asym or MNI152_2009c_sym? Are there any differences between asym version and sym version?
Secondly, the above errors were found when I applied 3dNetCorr to surface data. 3dNetCorr can not handle the two hemisphere data as a whole at present so if you want to get a whole-brain correlation matrix, you have to extract all roi stats from two hemispheres and do correlations analysis using other tools manually. That is cumbersome.