theodpzz/ct-scroll

[MIDL 2025] Imitating Radiological Scrolling: A Global-Local Attention Model for 3D Chest CT Volumes Multi-Label Anomaly Classification 🩺👨🏻‍⚕️

✅ PyTorch pretrained model weights of"Imitating Radiological Scrolling: A Global-Local Attention Model for 3D Chest CT Volumes Multi-Label Anomaly Classification".

📄 Accepted at MIDL 2025: arXiv preprint.

⚡️ PyTorch implementation available at https://github.com/theodpzz/ct-scroll.

🔥 Available resources

./ckpt/model_state_dict.pt: Model weights for CT-SSG trained on the CT-RATE training set.

./ckpt/thresholds.json: Per-abnormality classification thresholds optimized on our internal CT-RATE validation set. The official CT-RATE test set was not used during threshold optimization to preserve unbiased evaluation.

🤝🏻 Acknowledgment

We thank contributors from the CT-RATE dataset available at https://huggingface.co/datasets/ibrahimhamamci/CT-RATE, and from the Rad-ChestCT dataset available at https://zenodo.org/records/6406114.

Purpose

The model, trained on a publicly available dataset, is provided for academic and research purposes only, to support reproducibility of the results described in the associated paper. This repository is a research prototype, and is not intended for clinical use.

📎Citation

If you find this repository useful for your work, we would appreciate the following citation:

@InProceedings{dipiazza_2025_ctscroll,
        title = {Imitating Radiological Scrolling: A Global-Local Attention Model for 3D Chest CT Volumes Multi-Label Anomaly Classification},
        author = {Di Piazza, Theo and Lazarus, Carole and Nempont, Olivier and Boussel, Loic},
        booktitle = {Proceedings of The 8nd International Conference on Medical Imaging with Deep Learning -- MIDL 2025},
        year = {2025},
        publisher = {PMLR},
}