Modeling Annotator Preference and Stochastic Annotation Error for Medical Image Segmentation (PADL)

This repo contains the official implementation of our paper: Modeling Annotator Preference and Stochastic Annotation Error for Medical Image Segmentation, which highlights the issue of annotator-related biases existed in medical image segmentation tasks.

Paper

Requirements

This repo was tested with Ubuntu 20.04.4 LTS, Python 3.8, PyTorch 1.7.1, and CUDA 10.1. We suggest using virtual env to configure the experimental environment.

1. Clone this repo:

git clone https://github.com/Merrical/PADL.git

2. Create experimental environment using virtual env:

virtualenv .env --python=3.8 # create
source .env/bin/activate # activate
pip install -r requirements.txt

Dataset

The dataset details and the download link can be found here.

Training

python main.py --dataset RIGA --rater_num 6 --phase train --net_arch PADL --loss_func bce --device_id 0 --loop 0

Inference

python main.py --dataset RIGA --rater_num 6 --phase test --net_arch PADL --loss_func bce --device_id 0 --loop 0

Bibtex

@article{Liao2023PADL,
  title   = {Modeling Annotator Preference and Stochastic Annotation Error for Medical Image Segmentation},
  author  = {Liao, Zehui and Hu, Shishuai and Xie, Yutong and Xia, Yong},
  journal = {Medical Image Analysis},
  year    = {2023}
}

Contact Us

If you have any questions, please contact us ( [email protected] ).