Function orientation

monai/transforms/spatial/functional.py:229–272

Functional implementation of changing the input image's orientation into the specified based on `spatial_ornt`. This function operates eagerly or lazily according to ``lazy`` (default ``False``). Args: img: data to be changed, assuming `img` is channel-first. origin

(img, original_affine, spatial_ornt, lazy, transform_info)

def orientation(img, original_affine, spatial_ornt, lazy, transform_info) -> torch.Tensor:
    """
    Functional implementation of changing the input image's orientation into the specified based on `spatial_ornt`.
    This function operates eagerly or lazily according to
    ``lazy`` (default ``False``).

    Args:
        img: data to be changed, assuming `img` is channel-first.
        original_affine: original affine of the input image.
        spatial_ornt: orientations of the spatial axes,
            see also https://nipy.org/nibabel/reference/nibabel.orientations.html
        lazy: a flag that indicates whether the operation should be performed lazily or not
        transform_info: a dictionary with the relevant information pertaining to an applied transform.
    """
    spatial_shape = img.peek_pending_shape() if isinstance(img, MetaTensor) else img.shape[1:]
    xform = nib.orientations.inv_ornt_aff(spatial_ornt, spatial_shape)
    img = convert_to_tensor(img, track_meta=get_track_meta())

    spatial_ornt[:, 0] += 1  # skip channel dim
    spatial_ornt = np.concatenate([np.array([[0, 1]]), spatial_ornt])
    axes = [ax for ax, flip in enumerate(spatial_ornt[:, 1]) if flip == -1]
    full_transpose = np.arange(len(spatial_shape) + 1)  # channel-first array
    full_transpose[: len(spatial_ornt)] = np.argsort(spatial_ornt[:, 0])
    extra_info = {"original_affine": original_affine}

    shape_np = convert_to_numpy(spatial_shape, wrap_sequence=True)
    shape_np = shape_np[[i - 1 for i in full_transpose if i > 0]]
    meta_info = TraceableTransform.track_transform_meta(
        img,
        sp_size=shape_np,
        affine=xform,
        extra_info=extra_info,
        orig_size=spatial_shape,
        transform_info=transform_info,
        lazy=lazy,
    )
    out = _maybe_new_metatensor(img)
    if lazy:
        return out.copy_meta_from(meta_info) if isinstance(out, MetaTensor) else meta_info  # type: ignore
    if axes:
        out = torch.flip(out, dims=axes)
    if not np.all(full_transpose == np.arange(len(out.shape))):
        out = out.permute(full_transpose.tolist())
    return out.copy_meta_from(meta_info) if isinstance(out, MetaTensor) else out  # type: ignore


def flip(img, sp_axes, lazy, transform_info):