README.md

metadata

license: mit
task_categories:
  - image-segmentation
tags:
  - medical
  - cancer-screening
  - pan-cancer

Pancancer Dataset

Paper | Code

This repository contains the dataset for the paper "Glance and Focus Reinforcement for Pan-cancer Screening", accepted at ICLR 2026.

GF-Screen is a Glance and Focus reinforcement learning framework for pan-cancer screening in large-scale CT scans. It employs a Glance model to localize diseased regions and a Focus model to precisely segment lesions, addressing the challenge of localizing diverse types of tiny lesions in large CT volumes.

Download Dataset

You can download the dataset using the Hugging Face CLI:

huggingface-cli download linshangmail/Pancancer --repo-type dataset --local-dir . --cache-dir ./cache

Data Preparation

Following the official repository structure, the data should be organized as follows:

./ # project root
├──data
    ├──imagesTr
    ├──labelsTr
    ├──external
    ├──...
├──jsons
├──models
├──...

Usage

Training

Training requires one 80G GPU:

bash GF_RL_Screen.sh

Evaluation

# Internal validation:
python --trained_pth $YOUR_MODEL_PATH val_GF_internal.py

# External validation:
python --trained_pth $YOUR_MODEL_PATH --dataset_name FLARE23 val_GF_external.py

# FLARE prediction:
python --trained_pth $YOUR_MODEL_PATH --test_data_path $YOUR_PATH_TO_FLARE_VALIDATION_IMAGES val_GF_internal.py

Citation

If you find this repo or dataset useful for your research, please consider citing the paper as follows:

@inproceedings{GF-Screen,
  title={Glance and Focus Reinforcement for Pan-cancer Screening},
  author={Wu, Linshan and Zhuang, Jiaxin and Chen, Hao},
  booktitle={ICLR},
  year={2026},
}

@article{ma2024flare,
  title={Unleashing the strengths of unlabelled data in deep learning-assisted pan-cancer abdominal organ quantification: the FLARE22 challenge},
  author={Ma, Jun and Zhang, Yao and Gu, Song and Ge, Cheng and Mae, Shihao and Young, Adamo and Zhu, Cheng and Yang, Xin and Meng, Kangkang and Huang, Ziyan and others},
  journal={The Lancet Digital Health},
  volume={6},
  number={11},
  pages={e815--e826},
  year={2024},
  publisher={Elsevier}
}