Related Experiment Video
Updated: Aug 9, 2026
06:45
Automated Joint Space Detection Improves Bone Segmentation Accuracy
Published on: November 28, 2025
Beyond the pelvic ring: benchmarking multi-bone segmentation for ROI-Guided fracture detection
Siam Tahsin Bhuiyan1, Rashedur Rahman1, Sefatul Wasi2
1Center for Computational & Data Sciences, Independent University, Bangladesh, 1229, Dhaka, Bangladesh.
Scientific Reports
|August 7, 2026
Summary
Accurate pelvic fracture detection is improved by a novel Multi-Bone Segmentation Method. This AI approach precisely isolates pelvic bones on X-rays, enhancing diagnostic accuracy and interpretability for critical injuries.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Radiology
Background:
- Pelvic fractures are severe injuries with high mortality rates.
- Conventional radiography struggles with accurate diagnosis due to anatomical complexity.
- Automated fracture detection needs improved preprocessing for reliability.
Purpose of the Study:
- To evaluate the impact of segmentation-guided preprocessing on pelvic fracture detection.
- To introduce and assess a Multi-Bone Segmentation Method for enhanced X-ray analysis.
- To improve the accuracy and interpretability of AI-based fracture detection systems.
Main Methods:
- Developed a transformer-enhanced U-Net for segmenting nine individual pelvic bones.
- Compared the Multi-Bone Segmentation Method against conventional and ROI-guided approaches.
- Validated performance on PXR150 and AMERI PXR datasets, benchmarking against manual annotations.
Main Results:
- The Multi-Bone Segmentation Method significantly improved Accuracy and AUROC compared to other methods.
- Achieved diagnostic precision close to the manual reference standard by filtering artifacts.
- GradCAM visualizations confirmed alignment of model activation with fracture sites, enhancing interpretability.
Conclusions:
- Anatomical completeness in segmentation is crucial for reliable automated fracture detection.
- The Multi-Bone Segmentation Method offers a significant advancement in diagnosing pelvic fractures from X-rays.
- This approach enhances both the accuracy and explainability of AI diagnostic tools.
