Lambda transforms

lambda_transforms

ImageFunction

ImageFunction(
    function: typing.Callable[..., SimpleITK.Image],
    copy_geometry: bool = True,
    **kwargs: typing.Any
)

Bases: imgtools.transforms.base_transform.BaseTransform

Apply a custom function to process an image.

This class wraps a user-defined function and allows it to be used like any other transform in the library. The function must accept a SimpleITK image as its first parameter and return a SimpleITK image.

Parameters:

Name	Type	Description	Default
function	typing.Callable[..., SimpleITK.Image]	A callable that processes a SimpleITK image and returns a processed image.	required
copy_geometry	bool	If True, copies the input image's geometry to the result.	True
**kwargs	typing.Any	Optional keyword arguments to be passed to the processing function.	{}

Examples:

>>> import SimpleITK as sitk
>>> import numpy as np
>>> from imgtools.datasets.sample_images import (
...     create_sphere_image,
... )
>>> from imgtools.transforms import ImageFunction
>>> # Create a sample sphere image
>>> image = create_sphere_image()
>>> # Define a custom function to invert intensities
>>> def invert_intensity(image, max_value=1.0):
...     return sitk.Subtract(max_value, image)
>>> # Create a transform using this function
>>> inverter = ImageFunction(
...     invert_intensity, max_value=1.0
... )
>>> inverted = inverter(image)
>>> # Define a function that operates on the NumPy array
>>> def add_random_noise(image, noise_level=0.1):
...     array = sitk.GetArrayFromImage(image)
...     noise = np.random.normal(
...         0, noise_level, array.shape
...     )
...     array = array + noise
...     result = sitk.GetImageFromArray(array)
...     result.CopyInformation(image)
...     return result
>>> # Create and apply a noise transform
>>> noiser = ImageFunction(
...     add_random_noise, noise_level=0.2
... )
>>> noisy = noiser(image)
>>> print("Applied custom image transforms")
Applied custom image transforms

Registers a user-defined function for image processing along with its settings. The supplied function must accept a SimpleITK image and return a processed image. If copy_geometry is True, the geometry of the input image will be copied to the output.

Parameters:

Name	Type	Description	Default
function	typing.Callable[..., SimpleITK.Image]	A callable that processes a SimpleITK image and returns a processed image.	required
copy_geometry	bool	If True, copies the input image's geometry to the result.	True
**kwargs	typing.Any	Optional keyword arguments to be passed to the processing function.	{}

Source code in src/imgtools/transforms/lambda_transforms.py

def __init__(
    self,
    function: Callable[..., sitk.Image],
    copy_geometry: bool = True,
    **kwargs: Any,  # noqa: ANN401
) -> None:
    """Initialize an ImageFunction transform.

    Registers a user-defined function for image processing along with its
    settings. The supplied function must accept a SimpleITK image and
    return a processed image. If copy_geometry is True, the geometry of
    the input image will be copied to the output.

    Parameters
    ----------
    function : Callable[..., sitk.Image]
        A callable that processes a SimpleITK image and returns a
        processed image.
    copy_geometry : bool, optional
        If True, copies the input image's geometry to the result.
    **kwargs : Any, optional
        Optional keyword arguments to be passed to the processing function.
    """
    self.function = function
    self.copy_geometry = copy_geometry
    self.kwargs = kwargs

name property

name: str

Return the name of the transform class for logging and debugging.

Returns:

Type	Description
str	The name of the transform class.

SimpleITKFilter

SimpleITKFilter(
    sitk_filter: SimpleITK.ImageFilter,
    *execute_args: typing.Any
)

Bases: imgtools.transforms.base_transform.BaseTransform

Apply a SimpleITK image filter to an image.

This class wraps a SimpleITK image filter and allows it to be used like any other transform in the library. It provides a consistent interface for applying SimpleITK filters within the transform framework.

Parameters:

Name	Type	Description	Default
sitk_filter	SimpleITK.ImageFilter	A SimpleITK image filter instance to process images.	required
*execute_args	typing.Any	Optional positional arguments to pass to the filter's Execute method. These arguments come after the input image in the Execute call.	()

Examples:

>>> import SimpleITK as sitk
>>> from imgtools.datasets.sample_images import (
...     create_noisy_sphere_image,
... )
>>> from imgtools.transforms import SimpleITKFilter
>>> # Create a sample image with noise
>>> image = create_noisy_sphere_image(
...     noise_level=0.3
... )
>>> # Create a median filter to remove noise
>>> median_filter = sitk.MedianImageFilter()
>>> median_filter.SetRadius(2)
>>> # Use SimpleITKFilter to apply the filter
>>> median_transform = SimpleITKFilter(median_filter)
>>> filtered_image = median_transform(image)
>>> # Use SimpleITKFilter with a different filter that takes parameters
>>> gradient_magnitude = sitk.GradientMagnitudeRecursiveGaussianImageFilter()
>>> gradient_transform = SimpleITKFilter(
...     gradient_magnitude, 1.5
... )  # sigma=1.5
>>> gradient_image = gradient_transform(image)
>>> print(
...     "Applied median filter and gradient magnitude filter"
... )
Applied median filter and gradient magnitude filter

Parameters:

Name	Type	Description	Default
sitk_filter	SimpleITK.ImageFilter	A SimpleITK image filter instance to process images.	required
*execute_args	typing.Any	Optional positional arguments to pass to the filter's Execute method.	()

Source code in src/imgtools/transforms/lambda_transforms.py

def __init__(
    self,
    sitk_filter: sitk.ImageFilter,
    *execute_args: Any,  # noqa: ANN401
) -> None:
    """Initialize a SimpleITKFilter with a filter and execution args.

    Parameters
    ----------
    sitk_filter : sitk.ImageFilter
        A SimpleITK image filter instance to process images.
    *execute_args : Any, optional
        Optional positional arguments to pass to the filter's Execute
        method.
    """
    self.sitk_filter = sitk_filter
    self.execute_args = execute_args

name property

name: str

Return the name of the transform class for logging and debugging.

Returns:

Type	Description
str	The name of the transform class.