SaveImage doesn't work well

[paproaskwar]

I am trying to use Autoencoder reference to reconstruct the input CT image. I want to change the autoencoder's output's space, so I use the resample and set the meta equals to the input's meta
the code is:

reconstruction, _, _ = dynamic_infer(val_inferer, autoencoder, images)
reconstruction.meta = images.meta
img_saver = SaveImage(
output_dir="output",
output_postfix=output_postfix + "_image",
output_ext=".nii.gz",
separate_folder=False,
resample=True,
mode="nearest",
# dtype=np.float32
)
img_saver(reconstruction[0]) # image is the input CT

However it can't work:

Only 2d and 3d affine are supported, got 0d input.
Traceback (most recent call last):
File ".../maisi/test/xu_ttest.py", line 133, in
img_saver(reconstruction[0])
File "../enter/envs/new/lib/python3.9/site-packages/monai/transforms/io/array.py", line 520, in call
raise RuntimeError(
RuntimeError: SaveImage cannot find a suitable writer for output/sample_20250108_210439_751563_image.nii.gz.
Please install the writer libraries, see also the installation instructions:
https://docs.monai.io/en/latest/installation.html#installing-the-recommended-dependencies.
The current registered writers for .nii.gz: (<class 'monai.data.image_writer.NibabelWriter'>, <class 'monai.data.image_writer.ITKWriter'>).
Traceback (most recent call last):
File "../enter/envs/new/lib/python3.9/site-packages/monai/transforms/io/array.py", line 505, in call
writer_obj.set_metadata(meta_dict=meta_data, **self.meta_kwargs)
File "../enter/envs/new/lib/python3.9/site-packages/monai/data/image_writer.py", line 604, in set_metadata
self.data_obj, self.affine = self.resample_if_needed(
File "../enter/envs/new/lib/python3.9/site-packages/monai/data/image_writer.py", line 273, in resample_if_needed
output_array = resampler(
File "../enter/envs/new/lib/python3.9/site-packages/monai/transforms/spatial/array.py", line 226, in call
return spatial_resample(
File "../enter/envs/new/lib/python3.9/site-packages/monai/transforms/spatial/functional.py", line 105, in spatial_resample
src_affine = to_affine_nd(spatial_rank, src_affine).to(torch.float64)
File "../enter/envs/new/lib/python3.9/site-packages/monai/data/utils.py", line 977, in to_affine_nd
raise ValueError(f"affine must have 2 dimensions, got {affine_np.ndim}.")
ValueError: affine must have 2 dimensions, got 3.

Traceback (most recent call last):
File ".../enter/envs/new/lib/python3.9/site-packages/monai/transforms/io/array.py", line 505, in call
writer_obj.set_metadata(meta_dict=meta_data, **self.meta_kwargs)
File "../enter/envs/new/lib/python3.9/site-packages/monai/data/image_writer.py", line 440, in set_metadata
self.data_obj, self.affine = self.resample_if_needed(
File "../enter/envs/new/lib/python3.9/site-packages/monai/data/image_writer.py", line 273, in resample_if_needed
output_array = resampler(
File "../enter/envs/new/lib/python3.9/site-packages/monai/transforms/spatial/array.py", line 226, in call
return spatial_resample(
File "../enter/envs/new/lib/python3.9/site-packages/monai/transforms/spatial/functional.py", line 105, in spatial_resample
src_affine = to_affine_nd(spatial_rank, src_affine).to(torch.float64)
File "../enter/envs/new/lib/python3.9/site-packages/monai/data/utils.py", line 977, in to_affine_nd
raise ValueError(f"affine must have 2 dimensions, got {affine_np.ndim}.")
ValueError: affine must have 2 dimensions, got 3.

[paproaskwar]

I got the reason. When I read the input images, using EnsureChannelFirstd, which changes the affine from 2d to 3d