MSAN-CNN

This is a PyTorch implementation of the paper 'Weakly-supervised framework for cancer regions detection of hepatocellular carcinoma in whole-slide pathological images based on multi-scale attention convolutional neural network', and We'll refine the data and code over time.

🤝 Authorization

If you would like to use our data, please contact us first and obtain authorization to use it.

🤝 Citation

If you find this code is useful for your research, please consider citing:

@article{
title={Weakly-supervised framework for cancer regions detection of hepatocellular carcinoma in whole-slide pathological images based on multi-scale attention convolutional neural network}，
author={Songhui Diao, Yinli Tian, Wanming Hu, Jiaxin Hou, Ricardo Lambo, Zhicheng Zhang, Yaoqin Xie, Xiu Nie, Fa Zhang, Racoceanu Daniel, Wenjian Qin}，
journal={XXX}，
year={2021},
}

Setup

Prerequisites

• PyTorch 1.9.0
• python 3.8.5
• Torchvision 0.10.0
• numpy and so on

Data

• train data
• validation data
• test data

Training

Training a network with default arguments. Model checkpoints and tensorboard logs are written out to a unique directory created by default within experiments/models and experiments/logs respectively after starting training. If conditions permit, it will be better to pre-train the model first.

python main.py