vison307/pistoseg

Code Repository for AAAI23 paper "Weakly-Supervised Semantic Segmentation for Histopathology Images Based on Dataset Synthesis and Feature Consistency Constraint"

PistoSeg

Code Repository for AAAI23 Paper "Weakly-Supervised Semantic Segmentation for Histopathology Images Based on Dataset Synthesis and Feature Consistency Constraint"

If you are from China Mainland, you can refer to the gitee repo https://gitee.com/vison307/PistoSeg.

Installation

Code tested on

• Ubuntu 18.04
• A single Nvidia GeForce RTX 3090
• Python 3.8
• Pytorch 1.12.1
• Pytorch Lightning 1.7.1

Please use the following command to install the dependencies:

conda env create -f environment.yaml

Preparing the Data and Weights

1. Download the WSSS4LUAD dataset and put it in ./data/WSSS4LUAD

2. Download the BCSS-WSSS dataset and put it in ./data/BCSS-WSSS (Thanks to Han et. al)

3. Download the ImageNet-pretrained ResNet weight ilsvrc-cls_rna-a1_cls1000_ep-0001.params from SEAM repository and put it in ./weights/ilsvrc-cls_rna-a1_cls1000_ep-0001.params

4. Download the CAM model's weights res38d.pth from OEEM repository and put it in ./weights/res38d.pth

CAM Generation

Please refer to the OEEM readme in ./OEEM/README.md

Or you can use the pre-generated CAMs (in AAAI23/{dataset}/data/CAM/train.zip) and unzip and put them in ./data/WSSS4LUAD/CAM/train and ./data/BCSS-WSSS/CAM/train.

Train & Test

WSSS4LUAD

1. Assure that the CAM of the training set is put into ./data/WSSS4LUAD/CAM/train (in .npy format)

2. Split the validation and test dataset with split_validation.ipynb to produce 224x224 regular-shaped patches

3. Prepare the synthesized dataset and background mask with create_dataset.ipynb

4. Run bash run.sh

BCSS-WSSS

1. Assure that the CAM of the training set is put into ./data/BCSS-WSSS/CAM/train (in .npy format)

2. Prepare the synthesized dataset with create_dataset_bcss.ipynb

3. run bash run-bcss.sh

Note

Due to a disaster to the server, the original weights of the results provided in the paper are lost. So the current codes and weights are re-implemented and re-trained. The results are a bit different from the original ones (but slightly improved in mIoU). The overall performance is still similar.

WSSS4LUAD

• Test mIoU: 0.7530
• Test fwIoU: 0.7582
• Test tissue IoU: [0.7991 0.7020 0.7580] - [TUM, STR, NOM]

BCSS-WSSS

• Test mIoU: 0.7075
• Test fwIoU: 0.7576
• Test tissue IoU: [0.8144 0.7446 0.6063 0.6645] - [TUM, STR, LYM, NEC]

Reproducibility

We tried our best to ensure the reproducibility of the results, but since the torch.nn.functional.interpolate function is not deterministic, the results may be different over runs. If you want to fully reproduce the results, you can use the following weights (code: yj84) (preliminary segmentation epoch=*.ckpt, refining ResNet38-RFM.pth, and precise segmentation segmentation_log/epoch=*.ckpt) and intermediate results (Generated CAM data/CAM/train.zip and Refined Masks refine/CAM.zip). Training logs are also provided for reference.

Citation

If you find our code helpful, please cite as follows:

@inproceedings{fang2023weakly, title={Weakly-supervised semantic segmentation for histopathology images based on dataset synthesis and feature consistency constraint}, author={Fang, Zijie and Chen, Yang and Wang, Yifeng and Wang, Zhi and Ji, Xiangyang and Zhang, Yongbing}, booktitle={Proceedings of the AAAI Conference on Artificial Intelligence}, volume={37}, number={1}, pages={606--613}, year={2023} }

pistoseg's Issues

Split dataset

Dear author：
具体如何使用split_validation.ipynb来Split the validation and test dataset呢？

Where can I find Supplemental Materials of this paper?

I have only found the 8 pages of the main text of this paper and could not locate the Supplemental Materials. Could you please share the link of the Supplemental Materials if it is available? Thank you very much!

Weights for OEEM

Hello, I would like to reproduce your work, and tried to download CAM model's weights from OEEM repository. But I cannot download it from there. So, can you upload its weight on google drive or something I can access?

能在这里发一下论文的全文链接吗，我想看看论文，但是百度好像没搜索到。

对比实验请教

哈喽，这是一项非常有趣且有意义的工作，恭喜你们的论文在AAAI2023上发布！我注意到在论文的对比实验中你们使用了C-CAM 2022 CVPR的实验，目前我同样有工作需要和其对比但是数据集不同并且我在网上并没有找到相关的开源代码，请问你是如何得到这项实验结果的呢？

哈喽，AAAI23是结束了吗？代码整理工作咋样了

首先恭喜你的论文选入AAAI23大会，其次您论文的想法我觉得新奇且具有一定潜力，非常期待相关在相关工作中引入你的**，作为实验一个基线数据。看看是否有提升。

我在WSSS4LUAD的数据集挑战赛中看到更高的分数，评价规则也是mIOU

他们的分数已经达到了0.84，请问相比于论文实验使用对比数据中的mIOU相差较大，可以解释一下吗，这个是挑战链接：https://wsss4luad.grand-challenge.org/，排行榜中有得分数据。