histox-contrib 0.1.2

Community contributions for HisToX

histox-contrib

Community-contributed GPL-3 extensions for HistoX.

This package provides additional deep learning tools for HistoX that are licensed under the GNU General Public License v3 (GPL-3). It is distributed separately from the main HistoX package to keep the core library free from GPL restrictions.

Based on slideflow-gpl by James Dolezal, adapted for HistoX.

---

Installation

pip install histox-contrib

Or install from source:

git clone https://github.com/你的用户名/histox-contrib.git
cd histox-contrib
pip install -e .

Note: histox must be installed first.

pip install histox

---

Features

Module	Description
RetCCL	Pretrained feature extractor (RetCCL)
CTransPath	Pretrained feature extractor (CTransPath)
CLAM	Clustering-constrained Attention Multiple Instance Learning

---

Usage

Once installed, histox-contrib automatically registers its components into HistoX via the plugin system. No manual import is required.

import histox as hx

# RetCCL and CTransPath are available after histox-contrib is installed
extractor = hx.build_feature_extractor('retccl')

---

License

This project is licensed under the GNU General Public License v3.0. See LICENSE for details.

This package is based on slideflow-gpl (Copyright 2024 James Dolezal), modified and adapted for HistoX.

---

Citation

If you use this package in your research, please cite the original works:

RetCCL:

@article{wang2023retccl,
  title={RetCCL: Clustering-guided contrastive learning for whole-slide image retrieval},
  author={Wang, Xiyue and others},
  journal={Medical Image Analysis},
  year={2023}
}

CTransPath:

@article{wang2022ctranspath,
  title={Transformer-based unsupervised contrastive learning for histopathological image classification},
  author={Wang, Xiyue and others},
  journal={Medical Image Analysis},
  year={2022}
}

CLAM:

@article{lu2021clam,
  title={Data-efficient and weakly supervised computational pathology on whole-slide images},
  author={Lu, Ming Y and others},
  journal={Nature Biomedical Engineering},
  year={2021}
}