TAIX-Ray Code

The official codebase for the TAIX-Ray paper.

Please see our paper for a detailed description: TAIX-Ray Paper

Setup

1. Clone the Repository

git clone https://github.com/TruhnLab/TAIX-Ray.git
cd TAIX-Ray

2. Install Dependencies

Create a conda environment and install the required dependencies:

conda env create -f environment.yaml
conda activate taix-ray

Dataset Preparation

1. Download the Dataset

The dataset can be downloaded from Hugging Face: TAIX-Ray Dataset

2. Set Dataset Path

Update the dataset path in the following file:

cxr/data/datasets/cxr_dataset.py

3. Verify the Dataset

Run the dataset verification script:

python tests/data/test_dataset.py

Training (Optional)

Train models using the following commands:

1. Train the Binary Classification Model

python scripts/main_train.py --task binary --model MST

2. Train the Ordinal Classification Model

python scripts/main_train.py --task ordinal --model MST --regression

Evaluation

1. Download Pretrained Model Weights

Pretrained model checkpoints can be downloaded from: TAIX-Ray Models

2. Evaluate the Binary Classification Model

python scripts/main_predict_binary.py --path_run path/to/checkpoint.ckpt

3. Evaluate the Ordinal Classification Model

python scripts/main_predict_ordinal.py --path_run path/to/checkpoint.ckpt

Citation

If you use this work in your research, please cite:

@article{truhn_comprehensive_2026,
	title = {A comprehensive bedside chest radiography dataset with structured, itemized and graded radiologic reports},
	volume = {13},
	issn = {2052-4463},
	url = {https://www.nature.com/articles/s41597-026-07271-7},
	doi = {10.1038/s41597-026-07271-7},
	journal = {Scientific Data},
	author = {Truhn, Daniel and Geiger, Daniel and Siepmann, Robert and Von Der Stück, Marc Sebastian and Bressem, Keno Kyrill and Kather, Jakob Nikolas and Kuhl, Christiane and Müller-Franzes, Gustav and Nebelung, Sven},
	year = {2026},
	pages = {632},
}