STAS(micropapillary, solid nests, single cells)

Liangrui Pan et al. is a developer helper.

English · 简体中文

Use the pre-trained model for feature preprocessing and build the spatial topology of WSI.

Features extracted based on KimiaNet and CTransPath. Please refer to KimiaNet: https://github.com/KimiaLabMayo/KimiaNet. Please refer to CTransPath: https://github.com/Xiyue-Wang/TransPath

python new_cut7.py

python new_cut7-1.py

Use KNN (K=9) to construct the spatial topology map.

python construct_graph1.py

train_feature1.py

• Best model for five-fold cross validation

  link：https://pan.baidu.com/s/11dxmND9ZhEA-o-Hnql6_rg?pwd=l6gh 

• Best model finally tested

  link：https://pan.baidu.com/s/1lT8x_ovemj3FXvfjTRjxmA?pwd=516i  

• Test the model to obtain predictions.

  python test_stas.py

• Only features of the histopathology image data are provided as the data has a privacy protection agreement.

  link：https://pan.baidu.com/s/1pJY1Cv9d-ML7jU09RnGOjg?pwd=rzj7
  https://zenodo.org/records/11611418

• We provide clinical data on STAS patients, including patient age, gender, stage and protein level expression data. Please contact the corresponding author or first author by email.

We welcome you to visit our STAS test platform at http://plr.20210706.xyz:5000/.

Clear all before use
1.Upload the pathology image of .svs file.
2.Click Submit to get the prediction results.

The code will be updated after the paper is accepted!! License MIT