MNI Registration changes real volumes

[qwang0225]

Hi, I have been exploring MRI registration recently. After registering a Nifti format MRI image on MNI152 template, I found out that real total volumes of the image increased, which is not supposed to happen to my understanding. The real volume = total number of voxels * real volume of each voxel. Please see the code below.

fixed_image = ants.image_read(template_path)
moving_image = ants.image_read(data_path)
transform = ants.registration(fixed_image,moving_image,'Affine')

reg3t = ants.apply_transforms(fixed_image,moving_image,transform['fwdtransforms'][0])
ants.image_write(reg3t, output_path)

mov_nii = nib.load(data_path)
out_nii = nib.load(output_path)


print(mov_nii.shape, out_nii.shape)
print(mov_nii.header.get_zooms(), out_nii.header.get_zooms())
print(np.prod(mov_nii.header.get_zooms()) * np.prod(mov_nii.shape), np.prod(out_nii.header.get_zooms()) * np.prod(out_nii.shape))

Here is the output.

(176, 198, 160)        (181, 217, 181)
(0.937, 0.937, 0.937)     (1.0, 1.0, 1.0)
4586871.7        7109137.0

So clearly registration changed the image spacing, so the shape of image and unit voxel changed, but I would expect the total volume remain the same. The reason I'm asking is that I try to calculate a segmented volume after feeding the registered image into a binary segmentation model, by multiplying the number of positive voxels and the volume of unit pixel based on output mask. I want to make sure that this way I will get the real volume of the segmentation.

Can any of you help out? Where did I miss? Thanks.

[KumoLiu]

Hi @ qwang0225, I suggest you ask the question in the ants repo.
https://discourse.slicer.org/t/the-difference-of-methods-in-ants-registration/25236
https://github.com/ANTsX/ANTs/wiki/Anatomy-of-an-antsRegistration-call

BTW, you can refer to
https://github.com/Project-MONAI/tutorials/blob/main/3d_registration/learn2reg_oasis_unpaired_brain_mr.ipynb
https://github.com/Project-MONAI/tutorials/blob/main/3d_registration/learn2reg_nlst_paired_lung_ct.ipynb

Hope it helps, thanks!