how to create a pipleline with Variable pixdim

[lg2578]

I readed some examples about spacing transform.like
Spacingd(
keys=["image", "label"],
pixdim=(1.0, 1.0, 1.0),
mode=("bilinear", "nearest"),
),
I need to know a image's spacing first,and pixdim fixed in the transforms pipeline.
how to create a pipleline with Variable pixdim, because images put into transforms pipeline may have different spacing.
I find there is affine matrix when loadimage,still I don't know how to do it.

[Nic-Ma]

Hi @ericspod @dongyang0122 ,

Could you please help share some best practices for this question?

Thanks in advance.

[dongyang0122]

@lg2578 the transform Spacingd converts the volumes to ones with the target spacing (fixed spacing, (1.0, 1.0, 1.0) in your example). It can take volumes with various spacing as input. not sure if this is what you asked.

[lg2578]

@dongyang0122 @Nic-Ma ,I've know a little,maybe, if there are two images, one is 300300,and spacing=(1.5,1.5),the other is 400400,with spacing=(2,2), then set spacing to (1,1),will the two images output shape 200*200 ,it's right?
but when I tried it ,get some errors.
my code is here: https://www.kaggle.com/lvgang623/monai-test
code:

trans=T.Compose([
    T.LoadImage(),
    T.AddChannel(),
    T.Spacing(pixdim=(1,1,3)),
   
])
path='../input/3d-solution-with-monai-produce-3d-data/train/case101/case101_day22/case101_day22_image.nii.gz'
img=trans(path)
print(img[1]['pixdim'])
img[0].shape

error:

KeyError Traceback (most recent call last)
/kaggle/input/monai-v081/monai/transforms/transform.py in apply_transform(transform, data, map_items, unpack_items, log_stats)
87 if isinstance(data, (list, tuple)) and map_items:
---> 88 return [_apply_transform(transform, item, unpack_items) for item in data]
89 return _apply_transform(transform, data, unpack_items)

/kaggle/input/monai-v081/monai/transforms/transform.py in (.0)
87 if isinstance(data, (list, tuple)) and map_items:
---> 88 return [_apply_transform(transform, item, unpack_items) for item in data]
89 return _apply_transform(transform, data, unpack_items)

/kaggle/input/monai-v081/monai/transforms/transform.py in _apply_transform(transform, parameters, unpack_parameters)
52
---> 53 return transform(parameters)
54

/kaggle/input/monai-v081/monai/transforms/utility/array.py in call(self, img)
187 """
--> 188 return img[None]
189

KeyError: None

The above exception was the direct cause of the following exception:

RuntimeError Traceback (most recent call last)
/tmp/ipykernel_33/871096864.py in
6 ])
7 path='../input/3d-solution-with-monai-produce-3d-data/train/case101/case101_day22/case101_day22_image.nii.gz'
----> 8 img=trans(path)
9 print(img[1]['pixdim'])
10 img[0].shape

/kaggle/input/monai-v081/monai/transforms/compose.py in call(self, input_)
171 def call(self, input_):
172 for transform in self.transforms:
--> 173 input = apply_transform(transform, input, self.map_items, self.unpack_items, self.log_stats)
174 return input_
175

/kaggle/input/monai-v081/monai/transforms/transform.py in apply_transform(transform, data, map_items, unpack_items, log_stats)
111 else:
112 _log_stats(data=data)
--> 113 raise RuntimeError(f"applying transform {transform}") from e
114
115

RuntimeError: applying transform <monai.transforms.utility.array.AddChannel object at 0x7fe7c572a9d0>