panoptica 0.5.14

Panoptic Quality (PQ) computation for binary masks.

panoptica

Computing instance-wise segmentation quality metrics for 2D and 3D semantic- and instance segmentation maps.

Features

The package provides three core modules:

1. Instance Approximator: instance approximation algorithms to extract instances from semantic segmentation maps/model outputs.
2. Instance Matcher: matches predicted instances with reference instances.
3. Instance Evaluator: computes segmentation and detection quality metrics for pairs of predicted - and reference segmentation maps.

Installation

With a Python 3.10+ environment, you can install panoptica from pypi.org:

pip install panoptica

Use cases and tutorials

For tutorials featuring various use cases, cf. BrainLesion/tutorials/panoptica.

Semantic Segmentation Input

Although an instance-wise evaluation is highly relevant and desirable for many biomedical segmentation problems, they are still addressed as semantic segmentation problems due to the lack of appropriate instance labels.

Jupyter Notebook Example

This tutorial leverages all three modules.

Unmatched Instances Input

It is a common issue that instance segmentation outputs feature good outlines but mismatched instance labels. For this case, modules 2 and 3 can be utilized to match the instances and report metrics.

Jupyter Notebook Example

Matched Instances Input

If your predicted instances already match the reference instances, you can directly compute metrics with the third module, see Jupyter Notebook Example](https://github.com/BrainLesion/tutorials/tree/main/panoptica/example_spine_matched_instance.ipynb)

Citation

If you use panoptica in your research, please cite it to support the development!

TBA

upcoming citation