using the Brainnetome atlas in SUMA


We are trying to view the Brainnetome atlas in SUMA and we are having difficulty.

We downloaded the Brainnetome atlas file from the afni website (, but it is unclear how to view that in SUMA and we also don’t know what anatomy (underlay) to use to view that file in SUMA. We tried to view it on the fasverage brain and also an MNI_152 brain, but they did not align. There were some parts of the cortex that were not covered by the atlas (see attached figure).

Can you please provide some guidance on how to view this atlas in SUMA?

Christine Smith

Screen Shot.png

How are you mapping the data to the surface? If you are using the default afni-suma communications, then you will only get the intersections of the surface with the voxels of the volume. That will often leave holes if the mask bends even just a little away from the volume. If you are doing this interactively, you can use the Vol2surf plugin to adjust how that’s done and allow for the mode/median/maximum along the segments from the pial to smoothwm surfaces. The mode usually makes the most sense, I think.

The Brainnetome atlas was based on the HCP-40 dataset, a templatized subset of the Human Connectome Project anatomical data. It’s aligned to 2009c dataset, but it is slightly different, so it doesn’t fill the edges perfectly either.

To fix both problems, you can use the interactive Vol2Surf plugin or use this script to project the atlas to the surface or download the niml.dset mapping here. Here I project with 3dVol2Surf using the non-zero mode, and then smooth the atlas on the surface with modal smoothing in a 5mm neighborhood. Find the links below for the atlas projection data and the MNI surfaces. The projected data is not perfectly filled and smooth, but it is usable.

Viewing the surface atlas, now in a surface mapping file that ends with .niml.dset, involves loading that dataset in suma.

  1. Start suma with directory of spec file for the MNI surfaces, here with both hemispheres
    suma -spec ~/suma_MNI152_2009/std.60.MNI152_2009_both.spec
    -sv ~/abin/MNI152_2009_template.nii.gz
  2. Right click on a surface and Ctrl-s to open the surface object controller.
  3. Load dset in the lower left part of that menu.
  4. Select the bnsurfatlas_beta0_mode5mm.${hemi}.niml.dset depending on which hemisphere surface is selected. The contralateral hemisphere will automatically be loaded.
  5. Change the colormap “Cmp” to be ROI_glasbey_256 or a similar colormap made for ROI indices.
  6. Adjust the shading by turning off background convexity coloring with the ‘b’ key over the suma image viewer/


project Brainnetome atlas to the MNI 152 2009c surface

this assumes the HCP40 is close enough to the 2009c asymmetric volume

where MNI152 2009c surfaces are

set surfpath = ~/suma_MNI152_2009

where the BN atlas is

set atlaspath = ~/abin

for both left and right hemispheres

foreach hemi ( lh rh )

project from the volume to the surface

use the non-zero mode to fill surface with something if any voxel along the path

has an atlas voxel

using the ld60 low resolution surface, but you can use ld141 with std.141…

3dVol2Surf -spec surfpath/std.60.MNI152_2009_both.spec \ -surf_A std.60.{hemi}.smoothwm.gii
-surf_B std.60.${hemi}.pial.gii -sv $atlaspath/MNI152_2009_template.nii.gz
-grid_parent $atlaspath/BN_Atlas_246_1mm.nii.gz -gp_index 0
-map_func nzmode -f_steps 10 -f_index nodes
-cmask "-a atlaspath/BN_Atlas_246_1mm.nii.gz -expr astep(a,0.000000)" \ -out_niml bnsurfatlas_beta0.{hemi}.niml.dset

this is a bonus set… the projection is quite noisy, so we can use the mode

again, but this time on the surface with a 5mm neighborhood

SurfLocalstat -hood 5 -stat mode -input bnsurfatlas_beta0.${hemi}.niml.dset
-i_gii surfpath/std.60.{hemi}.smoothwm.gii
-prefix bnsurfatlas_beta0_mode5mm.${hemi}.niml.dset

The MNI 2009c asymmetric surfaces are here:

While not exactly what you asked for, Brainnetome has a surface workflow that transforms atlas to the subject’s surface (