CCSeg: Costal Cartilage Segmentation Benchmark Dataset

We are excited to introduce CCSeg — the first publicly available benchmark dataset for Costal Cartilage Segmentation, designed to advance research in computer-aided diagnosis and surgical systems.

💡 Why CCSeg?

Costal cartilage segmentation presents significant challenges:

• Similar intensity values between foreground (cartilage) and background tissues (e.g., liver, intercostal muscles)
• Particularly difficult for adolescent patients due to softer cartilage texture
• Lack of public datasets and benchmarks severely limits research progress in this field

📊 Dataset Highlights

✅ 165 high-quality CT scan cases
✅ Precise voxel-level annotations (covering each individual costal cartilage)
✅ Multi-age group data (6-35 years old)
✅ Out-of-distribution (OOD) test set (22 cases) for generalization validation
✅ Multi-center data collection ensuring diversity

🏥 Data Source

• Primary data: Plastic Surgery Hospital, Chinese Academy of Medical Sciences (2014-2023)
• External test data: Second Hospital of Hebei Medical University (2021-2024)
• All data approved by ethics committees with informed patient consent

🔬 Professional Annotation Process

Segmentation was performed independently by 4 plastic surgery residents under the guidance of radiology experts, with final review and correction by senior plastic surgeons to ensure annotation accuracy and consistency.

📈 Dataset Split

• Training set: 85 cases
• Validation set: 40 cases
• Test set: 40 cases
• OOD test set: 22 cases

This benchmark dataset provides a solid foundation for medical image analysis research related to costal cartilage, with significant application value in plastic surgery procedures such as auricular reconstruction.

📥 Download & Resources

• Dataset: OSF | CCSeg
• Paper: Costal cartilage segmentation with topology guided deformable mamba: Method and benchmark
• Code: GitHub - EricwanAR/DeformableMambaSeg

📖 Citation

If you use this dataset or code in your research, please cite our paper:

@article{wang2025costal,
  title={Costal cartilage segmentation with topology guided deformable mamba: Method and benchmark},
  author={Wang, Senmao and Gong, Haifan and Cui, Runmeng and Wan, Boyao and Hu, Zhonglin and Yang, Haiqing and Zhou, Jingyang and Jiang, Haiyue and Lin, Lin},
  journal={Expert Systems with Applications},
  pages={130085},
  year={2025},
  publisher={Elsevier}
}