I can't believe ANTs is that good in Python

antspy.py

# Script written by Franco Matzkin (with the great help of this amazing tool)
# Install ANTsPy with:
# pip install antspyx
# (I'm not sure this is the last version).

import ants

moving_ct_path = 'imgs/CTimage.nii.gz'
moving_mask_path = 'imgs/CT_brain_mask.nii.gz'
fixed_ct_path = 'imgs/Atlas.nii.gz'
ct = ants.image_read(moving_ct_path)
mask = ants.image_read(moving_mask_path)
fixed = ants.image_read(fixed_ct_path)

# Register
my_tx = ants.registration(fixed=fixed, moving=ct,
                          type_of_transform='AffineFast',
                          default_value=-1500)
transf_ct = my_tx['warpedmovout']
reg_transform = my_tx['fwdtransforms']

# Save transform
ants.write_transform(ants.read_transform(reg_transform[0]), '/tmp/reg_ct.mat')

reg_ct_path = '/tmp/reg_ct.nii.gz'
ants.image_write(transf_ct, reg_ct_path)

# Apply the same transform to the mask
transf_mask = ants.apply_transforms(fixed=transf_ct, moving=mask,
                                    transformlist=reg_transform,
                                    interpolator='nearestNeighbor')

reg_mask_path = '/tmp/reg_mask.nii.gz'
ants.image_write(transf_mask, reg_mask_path)

# Go back to the original size
inverse_image = ants.apply_transforms(fixed=ct, moving=transf_ct,
                                      transformlist=reg_transform,
                                      interpolator='linear', imagetype=0,
                                      whichtoinvert=[True])

inverse_mask = ants.apply_transforms(fixed=ct, moving=transf_mask,
                                     transformlist=reg_transform,
                                     interpolator='nearestNeighbor',
                                     imagetype=0,
                                     whichtoinvert=[True])

inverse_image_path = '/tmp/inv_reg_ct.nii.gz'
ants.image_write(inverse_image, inverse_image_path)

inverse_mask_path = '/tmp/inv_reg_mask.nii.gz'
ants.image_write(inverse_mask, inverse_mask_path)

The 'default_value' seems to not have any effect on ANTsPy 0.2.7.