📌 News¶

✅ [April 7] The training datasets are now available for download.

✅ [May 14] Submission for the preliminary development phase is now open. Please check the guidelines to prepare for your contrainer submission.

✅ [July 10] Submission for the final test phase is now open. Deadline: August 16 (the exact local time can be found on the submission page).

✅ [August 17] To accommodate ongoing submissions, the organizers have extended the deadline by 72 hours. Participants who have already made their final submission will be allowed to update their submission.

✅ [August 30] Winners for the final test phase have been annouced. Congrats!

✅ [December 29 2025] Our challenge review paper has been accepted by IEEE TMI 🐧 (IEEE Early Access) (arXiv).¶

🐧 Background¶

Pelvic fractures, typically resulting from high-energy traumas, are among the most severe injuries, characterized by a disability rate over 50% and a mortality rate over 13%, ranking them as the deadliest of all compound fractures. The complexity of pelvic anatomy, along with surrounding soft tissues, makes surgical interventions especially challenging. Recent years have seen a shift towards the use of robotic-assisted closed fracture reduction surgeries, which have shown improved surgical outcomes. Accurate segmentation of pelvic fractures is essential, serving as a critical step in trauma diagnosis and image-guided surgery. In 3D CT scans, fracture segmentation is crucial for fracture typing, pre-operative planning for fracture reduction, and screw fixation planning. For 2D X-ray images, segmentation plays a vital role in transferring the surgical plan to the operating room via registration, a key step for precise surgical navigation. 

📋 Tasks¶

¶

As a MICCAI 2024 challenge, the PENGWIN segmentation challenge is designed to advance the development of automated pelvic fracture segmentation techniques in both 3D CT scans (Task 1) and 2D X-ray images (Task 2), aiming to enhance their accuarcy and robustness. Our dataset comprises CT scans from 150 patients scheduled for pelvic reduction surgery, collected from multiple institutions using a variety of scanning equipment. This dataset represents a diverse range of patient cohorts and fracture types. Ground-truth segmentations for sacrum and hipbone fragments have been semi-automatically annotated and subsequently validated by medical experts. Furthermore, we have generated high-quality, realistic X-ray images and corresponding 2D labels from the CT data using the DeepDRR method, incorporating a range of virtual C-arm camera positions and surgical tools. 

TL;DR
Task 1: Pelvic fragment segmentation on 3D CT
Task 2: Pelvic fragment segmentation on 2D X-ray

📎 Citation¶

% challenge summary paper (early access)
@ARTICLE{TMIPENGWIN,
  author={Sang, Yudi and Liu, Yanzhen and Yibulayimu, Sutuke and Wang, Yunning and Killeen, Benjamin D. and Liu, Mingxu and Ku, Ping-Cheng and Johannsen, Ole and Gotkowski, Karol and Zenk, Maximilian and Maier-Hein, Klaus and Isensee, Fabian and Yue, Peiyan and Wang, Yi and Yu, Haidong and Pan, Zhaohong and He, Yutong and Liang, Xiaokun and Liu, Daiqi and Fan, Fuxin and Jurgas, Artur and Skalski, Andrzej and Ma, Yuxi and Yang, Jing and Płotka, Szymon and Litka, Rafał and Zhu, Gang and Song, Yingchun and Unberath, Mathias and Armand, Mehran and Ruan, Dan and Kevin Zhou, S. and Cao, Qiyong and Zhao, Chunpeng and Wu, Xinbao and Wang, Yu},
  journal={IEEE Transactions on Medical Imaging}, 
  title={Benchmark of Segmentation Techniques for Pelvic Fracture in CT and X-Ray: Summary of the PENGWIN 2024 Challenge}, 
  year={2026},
  volume={},
  number={},
  pages={1-1},
  doi={10.1109/TMI.2025.3650126}
}

% benchmark method
@article{LIUandYIBULAYIMU2025MEDIA,
  title = {Preoperative fracture reduction planning for image-guided pelvic trauma surgery: A comprehensive pipeline with learning},
  journal = {Medical Image Analysis},
  volume = {102},
  pages = {103506},
  year = {2025},
  issn = {1361-8415},
  doi = {https://doi.org/10.1016/j.media.2025.103506},
  url = {https://www.sciencedirect.com/science/article/pii/S1361841525000544},
  author = {Yanzhen Liu and Sutuke Yibulayimu and Yudi Sang and Gang Zhu and Chao Shi and Chendi Liang and Qiyong Cao and Chunpeng Zhao and Xinbao Wu and Yu Wang},
}

% PENWIN dataset
@article{liu2025automatic,
  title={Automatic pelvic fracture segmentation: a deep learning approach and benchmark dataset},
  author={Liu, Yanzhen and Yibulayimu, Sutuke and Zhu, Gang and Shi, Chao and Liang, Chendi and Zhao, Chunpeng and Wu, Xinbao and Sang, Yudi and Wang, Yu},
  journal={Frontiers in Medicine},
  volume={12},
  pages={1511487},
  year={2025},
  publisher={Frontiers Media SA}
}