Leverage nilearn.datasets for atlases + templates about atlasreader

[miykael]

That's a great idea (even though I'm very sad about corgi).

As an additional pointer, I have a script that uses nilearn to quickly resample an atlas to a specific voxel resolution. It goes something as follows:

from nibabel import load, Nifti1Image
from nilearn.image import resample_img
from nibabel.spaces import vox2out_vox

# Desired voxel resolution
res = [2, 2, 3]

# Load image
img = load('1.0mm_brain.nii.gz')

# Get output shape and affine
target_shape, target_affine = vox2out_vox(img, voxel_sizes=[res, res, res])

# Resample image
res_img = resample_img(img, target_affine, target_shape, clip=True)

# Save image in new resolution
res_img.to_filename('2.0mm_brain.nii.gz')

Not sure if this is useful for all atlases, but at least for some I think it's useful.