Adversarial Perturbations Improve Generalization of Confidence Prediction in Medical Image Segmentation

We introduce a straightforward adversarial training strategy that enhances the reliability of direct confidence prediction in medical image segmentation under realistic domain shifts.

MIDL 2025 Conference Paper

Installation

1. Clone the deployment branch of this repo (no code, only docker utils)

git clone --branch deploy --single-branch [email protected]:MedVisBonn/midl25.git

2. Build the image -

cd midl25/docker
bash build.sh

3. Create shared direotories for data and other files, adapt docker/run.sh accordingly and create a container

bash run.sh

• (optional) in the container, navigate to /root/workplace/repos/midl25/ and create results/, pre-trained/monai-unets/ and pre-trained/score-predictor/ directories.

Usage

All applications can be run from bash files in src/apps.

• To train a U-Net, adapt src/apps/train_unet.sh and run it.
• To train a score predictor, adapt src/apps/trai_score_predictor.sh and run it.

Further configurations can be found in src/configs/unet/monai_unet.yaml, src/configs/model/score_predictor.yaml and their respective trainer configs in src/configs/trainer/.

Data

We evaluate our approach using two datasets: the SAML Dataset and the MNMS-2 Dataset. To work with these datasets, adapt the paths in the configuration files in src/configs/data to match your local environment. Any pre-processing is handled by the respective classes in src/dataset.

Citation & License

TBA

Contact

For questions, reach out to: lennartz (ät) cs.uni-bonn.de