Class RandSpatialCrop

monai/transforms/croppad/array.py:558–628

Crop image with random size or specific size ROI. It can crop at a random position as center or at the image center. And allows to set the minimum and maximum size to limit the randomly generated ROI. Note: even `random_size=False`, if a dimension of the expected ROI size is larger tha

class RandSpatialCrop(Randomizable, Crop):
    """
    Crop image with random size or specific size ROI. It can crop at a random position as center
    or at the image center. And allows to set the minimum and maximum size to limit the randomly generated ROI.

    Note: even `random_size=False`, if a dimension of the expected ROI size is larger than the input image size,
    will not crop that dimension. So the cropped result may be smaller than the expected ROI, and the cropped results
    of several images may not have exactly the same shape.

    This transform is capable of lazy execution. See the :ref:`Lazy Resampling topic<lazy_resampling>`
    for more information.

    Args:
        roi_size: if `random_size` is True, it specifies the minimum crop region.
            if `random_size` is False, it specifies the expected ROI size to crop. e.g. [224, 224, 128]
            if a dimension of ROI size is larger than image size, will not crop that dimension of the image.
            If its components have non-positive values, the corresponding size of input image will be used.
            for example: if the spatial size of input data is [40, 40, 40] and `roi_size=[32, 64, -1]`,
            the spatial size of output data will be [32, 40, 40].
        max_roi_size: if `random_size` is True and `roi_size` specifies the min crop region size, `max_roi_size`
            can specify the max crop region size. if None, defaults to the input image size.
            if its components have non-positive values, the corresponding size of input image will be used.
        random_center: crop at random position as center or the image center.
        random_size: crop with random size or specific size ROI.
            if True, the actual size is sampled from `randint(roi_size, max_roi_size + 1)`.
        lazy: a flag to indicate whether this transform should execute lazily or not. Defaults to False.
    """

    def __init__(
        self,
        roi_size: Sequence[int] | int,
        max_roi_size: Sequence[int] | int | None = None,
        random_center: bool = True,
        random_size: bool = False,
        lazy: bool = False,
    ) -> None:
        super().__init__(lazy)
        self.roi_size = roi_size
        self.max_roi_size = max_roi_size
        self.random_center = random_center
        self.random_size = random_size
        self._size: Sequence[int] | None = None
        self._slices: tuple[slice, ...]

    def randomize(self, img_size: Sequence[int]) -> None:
        self._size = fall_back_tuple(self.roi_size, img_size)
        if self.random_size:
            max_size = img_size if self.max_roi_size is None else fall_back_tuple(self.max_roi_size, img_size)
            if any(i > j for i, j in zip(self._size, max_size)):
                raise ValueError(f"min ROI size: {self._size} is larger than max ROI size: {max_size}.")
            self._size = tuple(self.R.randint(low=self._size[i], high=max_size[i] + 1) for i in range(len(img_size)))
        if self.random_center:
            valid_size = get_valid_patch_size(img_size, self._size)
            self._slices = get_random_patch(img_size, valid_size, self.R)

    def __call__(self, img: torch.Tensor, randomize: bool = True, lazy: bool | None = None) -> torch.Tensor:  # type: ignore
        """
        Apply the transform to `img`, assuming `img` is channel-first and