bupt-ai-cz / Glomeruli-Instance-Segmentation

A deep learning-based approach for glomeruli instance segmentation from multi-stained renal biopsy pathologic images

This repo is the official implementation of our paper "A deep learning-based approach for glomeruli instance segmentation from multi-stained renal biopsy pathologic images".

Pre-train models

Pytorch version: 1.2.0

Snapshot images

WSIs

Training

Run python code/tools/train.py code/tools/cascade_mask_rcnn_r50_fpn_1x.py --work_dir="code/work_dirs"

This repo is developed based on mmdetection. Please refer to get_started.md for more details.

Please cite this paper, if it helps your research:

@article{jiang2021deep,
  title={A Deep Learning-Based Approach for Glomeruli Instance Segmentation from Multistained Renal Biopsy Pathologic Images},
  author={Jiang, Lei and Chen, Wenkai and Dong, Bao and Mei, Ke and Zhu, Chuang and Liu, Jun and Cai, Meishun and Yan, Yu and Wang, Gongwei and Zuo, Li and others},
  journal={The American Journal of Pathology},
  volume={191},
  number={8},
  pages={1431--1441},
  year={2021},
  publisher={Elsevier}
}

Contact

Wenkai Chen

• email: [e-mail], [e-mail]
• wechat: cwkyiyi

If you have any questions, you can contact me directly.