Function map_binary_to_indices

monai/transforms/utils.py:446–479

Compute the foreground and background of input label data, return the indices after fattening. For example: ``label = np.array([[[0, 1, 1], [1, 0, 1], [1, 1, 0]]])`` ``foreground indices = np.array([1, 2, 3, 5, 6, 7])`` and ``background indices = np.array([0, 4, 8])`` Args:

(
    label: NdarrayOrTensor, image: NdarrayOrTensor | None = None, image_threshold: float = 0.0
)

def map_binary_to_indices(
    label: NdarrayOrTensor, image: NdarrayOrTensor | None = None, image_threshold: float = 0.0
) -> tuple[NdarrayOrTensor, NdarrayOrTensor]:
    """
    Compute the foreground and background of input label data, return the indices after fattening.
    For example:
    ``label = np.array([[[0, 1, 1], [1, 0, 1], [1, 1, 0]]])``
    ``foreground indices = np.array([1, 2, 3, 5, 6, 7])`` and ``background indices = np.array([0, 4, 8])``

    Args:
        label: use the label data to get the foreground/background information.
        image: if image is not None, use ``label = 0 & image > image_threshold``
            to define background. so the output items will not map to all the voxels in the label.
        image_threshold: if enabled `image`, use ``image > image_threshold`` to
            determine the valid image content area and select background only in this area.
    """
    check_non_lazy_pending_ops(label, name="map_binary_to_indices")
    # Prepare fg/bg indices
    if label.shape[0] > 1:
        label = label[1:]  # for One-Hot format data, remove the background channel
    label_flat = ravel(any_np_pt(label, 0))  # in case label has multiple dimensions
    fg_indices = nonzero(label_flat)
    if image is not None:
        check_non_lazy_pending_ops(image, name="map_binary_to_indices")
        img_flat = ravel(any_np_pt(image > image_threshold, 0))
        img_flat, *_ = convert_to_dst_type(img_flat, label, dtype=bool)
        bg_indices = nonzero(img_flat & ~label_flat)
    else:
        bg_indices = nonzero(~label_flat)

    # no need to save the indices in GPU, otherwise, still need to move to CPU at runtime when crop by indices
    fg_indices, *_ = convert_data_type(fg_indices, device=torch.device("cpu"))
    bg_indices, *_ = convert_data_type(bg_indices, device=torch.device("cpu"))
    return fg_indices, bg_indices


def map_classes_to_indices(