loki2.utils.tools

Utility functions for Loki2.

This module provides various helper functions for image processing, dictionary manipulation, file handling, and other common operations used throughout the Loki2 framework.

Module Contents

loki2.utils.tools.get_bounding_box(img: numpy.ndarray) → List[int]

Get bounding box coordinate information.

Parameters:

img – Binary or labeled image array.

Returns:

Bounding box coordinates as [rmin, rmax, cmin, cmax].

Return type:

List[int]

loki2.utils.tools.remove_small_objects_cp(pred: cupy.ndarray, min_size: int = 64, connectivity: int = 1) → cupy.ndarray

Remove connected components smaller than the specified size.

This function is taken from skimage.morphology.remove_small_objects, but the warning is removed when a single label is provided. Uses CuPy for GPU acceleration.

Parameters:

• pred – Input labelled array (CuPy array).

• min_size – Minimum size of instance in output array. Defaults to 64.

• connectivity – The connectivity defining the neighborhood of a pixel. Defaults to 1.

Returns:

Output array with instances removed under min_size.

Return type:

cp.ndarray

Raises:

ValueError – If negative value labels are found in the input.

loki2.utils.tools.remove_small_objects(pred: numpy.ndarray, min_size: int = 64, connectivity: int = 1) → numpy.ndarray

Remove connected components smaller than the specified size.

This function is taken from skimage.morphology.remove_small_objects, but the warning is removed when a single label is provided. Uses NumPy/ SciPy for CPU processing.

Parameters:

• pred – Input labelled array (NumPy array).

• min_size – Minimum size of instance in output array. Defaults to 64.

• connectivity – The connectivity defining the neighborhood of a pixel. Defaults to 1.

Returns:

Output array with instances removed under min_size.

Return type:

np.ndarray

Raises:

ValueError – If negative value labels are found in the input.

loki2.utils.tools.unflatten_dict(d: dict, sep: str = '.') → dict

Unflatten a flattened dictionary (create a nested dictionary).

Parameters:

• d – Dictionary to be nested (with keys separated by separator).

• sep – Separator used in flattened keys. Defaults to ‘.’.

Returns:

Nested dictionary reconstructed from flattened keys.

Return type:

dict

Example

>>> d = {'a.b': 1, 'a.c': 2}
>>> unflatten_dict(d)
{'a': {'b': 1, 'c': 2}}

loki2.utils.tools.remap_label(pred: numpy.ndarray, by_size: bool = False) → numpy.ndarray

Rename all instance IDs so that IDs are contiguous (e.g., [0, 1, 2, 3]).

The ordering of instances (which one comes first) is preserved unless by_size=True, in which case instances will be reordered so that larger nuclei have smaller IDs.

Parameters:

• pred – The 2D array containing instances where each instance is marked by a non-zero integer.

• by_size – If True, reorder instances so that larger nuclei have smaller IDs. Defaults to False.

Returns:

Array with remapped contiguous instance IDs.

Return type:

np.ndarray