Open Hip Dysplasia (radoss-org) Release V2.0
Adam McArthur and Jacob Jaremko and Abhilash Rakkunedeth Hareendranathan

@article{zenodo_16785185,
title= {Open Hip Dysplasia (radoss-org) Release V2.0},
journal= {},
author= {Adam McArthur and Jacob Jaremko and Abhilash Rakkunedeth Hareendranathan},
year= {2025},
url= {https://zenodo.org/records/16785185},
doi= {10.5281/zenodo.16785185},
abstract= {Open dataset for developmental dysplasia of the hip (DDH), released by the radoss-org project (University of Alberta and MEDO.ai). Release V2.0 adds X-ray data and snapshot information to the earlier ultrasound release.

**Source paper:** McArthur A, Wichuk S, Burnside S, Kirby A, Scammon A, Sol D, Hareendranathan A, Jaremko JL. *Retuve: Automated Multi-Modality Analysis of Hip Dysplasia with Open Source AI.* arXiv:2504.06422 (2025), https://arxiv.org/abs/2504.06422

**Background:** Hip dysplasia is hard to diagnose early, screening methods are not standardized, and many AI studies cannot be reproduced because the data and code are not shared. Retuve is an open source framework for analyzing hip dysplasia from both ultrasound and X-ray images. It ships with open datasets of expert annotated images, pretrained models with training code and weights, and a Python API.

**What the dataset is for:** The images and labels train and evaluate the segmentation and landmark detection models in Retuve. From those outputs the software measures standard diagnostic values such as the alpha angle (ultrasound) and the acetabular index (X-ray). For ultrasound, the pipeline uses ilium/acetabulum and femoral head segmentations and derives five landmarks from them.

**Contents:** A single archive, radoss-org/open-hip-dysplasia-2.0.zip (1.27 GB). Per the paper, the initial ultrasound release collates 200 2D ultrasound images from Radiopaedia and Hong Kong Polytechnic University. The paper says labels for an open X-ray dataset from The University of Jordan are planned for later work, and that more data will be released over time.

**Code and docs:** https://github.com/radoss-org/open-hip-dysplasia (this release is tag 2.0) and https://github.com/radoss-org/retuve

**Licensing:** Listed here as CC BY-NC-SA 3.0 (Attribution, NonCommercial, ShareAlike) to be safe. Zenodo also lists CC BY 4.0 and Apache 2.0 for this record, and the paper says the software is Apache 2.0 but gives no dataset license. Check the repository for which terms cover which files.

Imported from Zenodo record 16785185.

**Reference:** McArthur et al., Retuve: Automated Multi-Modality Analysis of Hip Dysplasia with Open Source AI. 2025. https://doi.org/10.1016/j.simpa.2025.100791 (arXiv: https://arxiv.org/abs/2504.06422)},
files= {radoss-org/open-hip-dysplasia-2.0.zip},
keywords= {segmentation, ultrasound, Medical imaging, landmark detection, hip dysplasia, x-ray, developmental dysplasia of the hip},
license= {CC BY-NC-SA 3.0},
terms= {},
superseded= {}
}