Metadata-Version: 2.1
Name: lacss
Version: 0.1.2
Summary: Cell segmentation and tracking
License: MIT
Author: Ji Yu
Author-email: jyu@uchc.edu
Requires-Python: >=3.8,<3.11
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.8
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Requires-Dist: cloudpickle (>=2.2.1,<3.0.0)
Requires-Dist: flax (>=0.6.6,<0.7.0)
Requires-Dist: imageio (>=2.26.0,<3.0.0)
Requires-Dist: numpy (>=1.24.1,<2.0.0)
Requires-Dist: scikit-image (>=0.20.0,<0.21.0)
Requires-Dist: tensorflow-cpu (>=2.8.0,<3.0.0)
Requires-Dist: tqdm (>=4.65.0,<5.0.0)
Description-Content-Type: text/markdown

# LACSS

LACSS is a model for single-cell segmentation and cell-lineage tracking

Ref: https://www.nature.com/articles/s42003-023-04608-5

As a segmentation model, it can work similar to other instance segmentation models such as MaskRCNN. However, it also support end-to-end training with very weak supervisions: e.g (a) image-level segmentation, and (b) location-of-interests (LOIs). These annotatins are chosen because they can often be produced progammably using simple unsupervised algorithms from experimental data. Our goal is to build a streamlined annotation-training pipeline that requires no manual input from humans.

The segmentation model is used for down-stream cell-tracking task. The tracking logic is based on SMC (sequential Monte Carlo).

This particular version of LACSS is build on [Jax](https://github.com/google/jax) framework. Both the segmentation model and the tracking logic heavily utilize the composable transformation facility provided by JAX.




