Why does Affine() change the metadata of the input MetaTensor too?

[ntatsisk]

Hello,

I noticed that Affine transformation changes the affine matrix for both the input and output MetaTensor because of this shallow copy here: https://github.com/Project-MONAI/MONAI/blob/dev/monai/transforms/spatial/array.py#L2294. Here is an example:

import torch
from monai.transforms import Affine
from monai.data.meta_tensor import MetaTensor

input_mt = MetaTensor(torch.rand((1, 5, 5, 5)))

print("Before affine:")
print(input_mt.meta)

affine_transform = Affine(translate_params=[2, 2, 2])
output_mt, _ = affine_transform(input_mt)

print("\nAfter affine:")
print(input_mt.meta)
print(output_mt.meta)

With output:

Before affine:
{affine: tensor([[1., 0., 0., 0.],
        [0., 1., 0., 0.],
        [0., 0., 1., 0.],
        [0., 0., 0., 1.]], dtype=torch.float64), space: RAS}

After affine:
{affine: tensor([[1., 0., 0., 2.],
        [0., 1., 0., 2.],
        [0., 0., 1., 2.],
        [0., 0., 0., 1.]], dtype=torch.float64), space: RAS}
{affine: tensor([[1., 0., 0., 2.],
        [0., 1., 0., 2.],
        [0., 0., 1., 2.],
        [0., 0., 0., 1.]], dtype=torch.float64), space: RAS}

I cannot understand why. Originally, I thought it was a bug but maybe is this part of "tracking the metadata" process of the MetaTensor? Why is it necessary? Thanks.