Method __call__

monai/transforms/spatial/array.py:2940–2987

Args: img: shape must be (num_channels, H, W, D), spatial_size: specifying spatial 3D output image spatial size [h, w, d]. if `spatial_size` and `self.spatial_size` are not defined, or smaller than 1, the transform will use the spatial

(
        self,
        img: torch.Tensor,
        spatial_size: tuple[int, int, int] | int | None = None,
        mode: str | int | None = None,
        padding_mode: str | None = None,
        randomize: bool = True,
    )

        self.rand_affine_grid.randomize()

    def __call__(
        self,
        img: torch.Tensor,
        spatial_size: tuple[int, int, int] | int | None = None,
        mode: str | int | None = None,
        padding_mode: str | None = None,
        randomize: bool = True,
    ) -> torch.Tensor:
        """
        Args:
            img: shape must be (num_channels, H, W, D),
            spatial_size: specifying spatial 3D output image spatial size [h, w, d].
                if `spatial_size` and `self.spatial_size` are not defined, or smaller than 1,
                the transform will use the spatial size of `img`.
            mode: {``"bilinear"``, ``"nearest"``} or spline interpolation order 0-5 (integers).
                Interpolation mode to calculate output values. Defaults to ``self.mode``.
                See also: https://pytorch.org/docs/stable/generated/torch.nn.functional.grid_sample.html
                When it's an integer, the numpy (cpu tensor)/cupy (cuda tensor) backends will be used
                and the value represents the order of the spline interpolation.
                See also: https://docs.scipy.org/doc/scipy/reference/generated/scipy.ndimage.map_coordinates.html
            padding_mode: {``"zeros"``, ``"border"``, ``"reflection"``}
                Padding mode for outside grid values. Defaults to ``self.padding_mode``.
                See also: https://pytorch.org/docs/stable/generated/torch.nn.functional.grid_sample.html
                When `mode` is an integer, using numpy/cupy backends, this argument accepts
                {'reflect', 'grid-mirror', 'constant', 'grid-constant', 'nearest', 'mirror', 'grid-wrap', 'wrap'}.
                See also: https://docs.scipy.org/doc/scipy/reference/generated/scipy.ndimage.map_coordinates.html
            randomize: whether to execute `randomize()` function first, default to True.
        """
        sp_size = fall_back_tuple(self.spatial_size if spatial_size is None else spatial_size, img.shape[1:])
        if randomize:
            self.randomize(grid_size=sp_size)

        _device = img.device if isinstance(img, torch.Tensor) else self.device
        grid = create_grid(spatial_size=sp_size, device=_device, backend="torch")
        if self._do_transform:
            if self.rand_offset is None:
                raise RuntimeError("rand_offset is not initialized.")
            gaussian = GaussianFilter(3, self.sigma, 3.0).to(device=_device)
            offset = torch.as_tensor(self.rand_offset, device=_device).unsqueeze(0)
            grid[:3] += gaussian(offset)[0] * self.magnitude
            grid = self.rand_affine_grid(grid=grid)
        out: torch.Tensor = self.resampler(
            img,
            grid,  # type: ignore
            mode=mode if mode is not None else self.mode,
            padding_mode=padding_mode if padding_mode is not None else self.padding_mode,
        )
        return out


class GridDistortion(Transform):