Journal Article2026

Beyond the pelvic ring: benchmarking multi-bone segmentation for ROI-Guided fracture detection

Siam Tahsin Bhuiyan, Rashedur Rahman, Sefatul Wasi, Halima Khatun, Manosh Sur Chowdhury, Riyadul Islam, Samiul Karim Mazumder, Naomi Yagi, Syoji Kobashi, Saadia Binte Alam

Scientific Reports

Nature Publishing Group UK London, Vol. 16, Issue 1, pp. 24508, ISBN: 2045-2322

  • Good Health and Well-being — Relevance: 86%
DOI: 10.1038/s41598-026-47663-8URL

Abstract

Pelvic fractures are critical injuries associated with high mortality rates, yet accurate diagnosis using conventional radiography remains challenging due to complex anatomical overlaps. This study evaluates the impact of segmentation-guided preprocessing on fracture detection from pelvic X-rays by introducing a Multi-Bone Segmentation Method. Unlike traditional approaches that restrict analysis to the pelvic ring or treat the pelvis as a single region, our framework utilizes a transformer-enhanced U-Net to distinctly isolate nine individual pelvic bones, creating an anatomically precise input for downstream classification. For segmentation, the model was trained on PXR150 and evaluated on both PXR150 and externally on AMERI PXR, while classification models were trained and evaluated separately for each dataset. We benchmarked this approach against three methods: a Conventional (raw X-ray input) baseline, a Conventional ROI-Guided Method (binary segmentation), and a Reference Segmentation (manual annotation) to establish a performance upper bound. Using the public PXR150 and private AMERI PXR datasets, we demonstrated that segmentation-guided analysis consistently enhances diagnostic precision. The Proposed Multi-Bone Method outperformed the Conventional ROI-Guided Method, achieving higher Accuracy (81.30%, 95% CI: 79.70-82.90 and 83.20%, 95% CI: 78.74-87.66) and AUROC (0.822, 95% CI: 0.794-0.851 and 0.838, 95% CI: 0.792-0.884) across both datasets. By effectively filtering non-anatomical artifacts, the multi-bone strategy achieved performance metrics close to the manually annotated reference standard. GradCAM visualizations further confirmed that multi-bone segmentation aligns model activation with clinically relevant fracture sites, enhancing interpretability. These findings underscore that anatomical completeness in segmentation is a decisive factor in improving the reliability and explainability of automated fracture detection systems.

Keywords

  • pelvic fractures
  • Grad-CAM interpretability
  • automated fracture detection
  • pelvic radiography

CCDS Authors

Siam Tahsin Bhuiyan
Siam Tahsin Bhuiyan

Siam Tahsin Bhuiyan

  • Intern, CCDS

Bachelors

Independent University, Bangladesh

Published at

Scientific Reports

Md. Rashedur Rahman, D. Eng.
Md. Rashedur Rahman, D. Eng.

Md. Rashedur Rahman, D. Eng.

  • Co-Director, MIRA & NEST Wings

Assistant Professor

Department of Computer Science and Engineering

Independent University, Bangladesh

Published at

CHI, HCII, Scientific Reports, Cognitive Computation, Neural Computing and Applications

Halima Khatun
Halima Khatun

Halima Khatun

  • Research Assistant, CCDS

Bachelors

Independent University, Bangladesh

Published at

Scientific Reports

Riyadul Islam
Riyadul Islam

Riyadul Islam

  • Intern, CCDS

Bachelors

Daffodil International University

Published at

Scientific Reports

Saadia Binte Alam, PhD
Saadia Binte Alam, PhD

Saadia Binte Alam, PhD

  • Director, MIRA & WiSE Wings

Associate Professor

Department of Computer Science and Engineering

Independent University, Bangladesh

Published at

HCII, Scientific Reports, Cognitive Computation, Neural Computing and Applications

References

  1. 1.Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, Li Fei-Fei. (2009). ImageNet: A large-scale hierarchical image database. 2009 IEEE Conference on Computer Vision and Pattern Recognition, 248–255[10.1109/cvpr.2009.5206848]
  2. 2.Ramprasaath R. Selvaraju, Michael Cogswell, Abhishek Das, Ramakrishna Vedantam, Devi Parikh, Dhruv Batra. (2017). Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization. , 618–626[10.1109/iccv.2017.74]
  3. 3.Fabian Isensee, Paul F. Jaeger, Simon A. A. Kohl, Jens Petersen, Klaus H. Maier-Hein. (2020). nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nature Methods, 18(2), 203–211[10.1038/s41592-020-01008-z]
  4. 4.Jun Ma, Yuting He, Feifei Li, Lin Han, Chenyu You, Bo Wang. (2024). Segment anything in medical images. Nature Communications, 15(1), 654[10.1038/s41467-024-44824-z]
  5. 5.Jakob Wasserthal, Hanns-Christian Breit, Manfred T. Meyer, Maurice Pradella, Daniel Hinck, Alexander W. Sauter, Tobias Heye, Daniel T. Boll, Joshy Cyriac, Shan Yang, Michael Bach, Martin Segeroth. (2023). TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images. Radiology Artificial Intelligence, 5(5)[10.1148/ryai.230024]
  6. 6.Kostejn, Vit, Essus, Yamil, Abrahamson, Jenna, Vatsavai, Ranga Raju. (2025). U-Prithvi: Integrating a Foundation Model and U-Net for Enhanced Flood Inundation Mapping. Leibniz international proceedings in informatics[10.4230/lipics.giscience.2025.18]
  7. 7.Seok Won Chung, Seung Seog Han, Ji Whan Lee, Kyung-Soo Oh, Na Ra Kim, Jong Pil Yoon, Joon Yub Kim, Sung Hoon Moon, Jieun Kwon, Hyo-Jin Lee, Young-Min Noh, Youngjun Kim. (2018). Automated detection and classification of the proximal humerus fracture by using deep learning algorithm. Acta Orthopaedica, 89(4), 468–473[10.1080/17453674.2018.1453714]
  8. 8.Julia P Grimes, Patrice M Gregory, Helaine Noveck, Mark S Butler, Jeffrey L Carson. (2002). The effects of time-to-surgery on mortality and morbidity in patients following hip fracture. The American Journal of Medicine, 112(9), 702–709[10.1016/s0002-9343(02)01119-1]
  9. 9.Chi-Tung Cheng, Tsung-Ying Ho, Tao-Yi Lee, Chih-Chen Chang, Ching-Cheng Chou, Chih-Chi Chen, I-Fang Chung, Chien-Hung Liao. (2019). Application of a deep learning algorithm for detection and visualization of hip fractures on plain pelvic radiographs. European Radiology, 29(10), 5469–5477[10.1007/s00330-019-06167-y]
  10. 10.C. Hurson, A. Tansey, B. O’Donnchadha, P. Nicholson, J. Rice, J. McElwain. (2007). Rapid prototyping in the assessment, classification and preoperative planning of acetabular fractures. Injury, 38(10), 1158–1162[10.1016/j.injury.2007.05.020]
  11. 11.Rubina Sarki, Khandakar Ahmed, Hua Wang, Yanchun Zhang. (2020). Automated detection of mild and multi-class diabetic eye diseases using deep learning. Health Information Science and Systems, 8(1), 32[10.1007/s13755-020-00125-5]
  12. 12.Min Liu, Lanlan Hu, Ying Tang, Chu Wang, Yu He, Chunyan Zeng, Kun Lin, Zhizi He, Wujie Huo. (2022). A Deep Learning Method for Breast Cancer Classification in the Pathology Images. IEEE Journal of Biomedical and Health Informatics, 26(10), 5025–5032[10.1109/jbhi.2022.3187765]
  13. 13.Dalia Alzu’bi, Malak Abdullah, Ismail Hmeidi, Rami AlAzab, Maha Gharaibeh, Mwaffaq El-Heis, Khaled H. Almotairi, Agostino Forestiero, Ahmad MohdAziz Hussein, Laith Abualigah. (2022). Kidney Tumor Detection and Classification Based on Deep Learning Approaches: A New Dataset in CT Scans. Journal of Healthcare Engineering, 2022, 1–22[10.1155/2022/3861161]
  14. 14.Antonio Pinto, Daniela Berritto, Anna Russo, Federica Riccitiello, Martina Caruso, Maria Paola Belfiore, Vito Roberto Papapietro, Marina Carotti, Fabio Pinto, Andrea Giovagnoni, Luigia Romano, Roberto Grassi. (2018). Traumatic fractures in adults: missed diagnosis on plain radiographs in the Emergency Department.. PubMed, 89(1-S), 111–123[10.23750/abm.v89i1-s.7015]
  15. 15.Leonardo Tanzi, Enrico Vezzetti, Rodrigo Moreno, Alessandro Aprato, Andrea Audisio, Alessandro Massè. (2020). Hierarchical fracture classification of proximal femur X-Ray images using a multistage Deep Learning approach. European Journal of Radiology, 133, 109373[10.1016/j.ejrad.2020.109373]
  16. 16.Chi-Tung Cheng, Yirui Wang, Huan-Wu Chen, Po-Meng Hsiao, Chun-Nan Yeh, Chi-Hsun Hsieh, Shun Miao, Jing Xiao, Chien-Hung Liao, Le Lu. (2021). A scalable physician-level deep learning algorithm detects universal trauma on pelvic radiographs. Nature Communications, 12(1), 1066[10.1038/s41467-021-21311-3]
  17. 17.Mohamed A. Kassem, Soaad M. Naguib, Hanaa M. Hamza, Mostafa M. Fouda, Mohamed K. Saleh, Khalid M. Hosny. (2023). Explainable Transfer Learning‐Based Deep Learning Model for Pelvis Fracture Detection. International Journal of Intelligent Systems, 2023(1)[10.1155/2023/3281998]
  18. 18.Husham Abdelrahman, Ayman El-Menyar, Holger Keil, Abduljabbar Alhammoud, Syed Imran Ghouri, Elhadi Babikir, Mohammad Asim, Matthias Muenzberg, Hassan Al-Thani. (2020). Patterns, management, and outcomes of traumatic pelvic fracture: insights from a multicenter study. Journal of Orthopaedic Surgery and Research, 15(1), 249[10.1186/s13018-020-01772-w]
  19. 19.Liding Yao, Xiaojun Guan, Xiaowei Song, Yanbin Tan, Chun Wang, Chaohui Jin, Ming Chen, Huogen Wang, Minming Zhang. (2021). Rib fracture detection system based on deep learning. Scientific Reports, 11(1), 23513[10.1038/s41598-021-03002-7]
  20. 20.C. Arvieux, F. Thony, C. Broux, F.-X. Ageron, E. Rancurel, J. Abba, J.-L. Faucheron, J.-J. Rambeaud, J. Tonetti. (2012). Current management of severe pelvic and perineal trauma. Journal of Visceral Surgery, 149(4), e227–e238[10.1016/j.jviscsurg.2012.06.004]
  21. 21.Kazutoshi Ukai, Rashedur Rahman, Naomi Yagi, Keigo Hayashi, Akihiro Maruo, Hirotsugu Muratsu, Syoji Kobashi. (2021). Detecting pelvic fracture on 3D-CT using deep convolutional neural networks with multi-orientated slab images. Scientific Reports, 11(1), 11716[10.1038/s41598-021-91144-z]
  22. 22.Elizabeth R. Benjamin, Dominik A. Jakob, Lee Myers, Panagiotis Liasidis, Meghan Lewis, Yong Fu, Demetrios Demetriades. (2022). The trauma pelvic X-ray: Not all pelvic fractures are created equally. The American Journal of Surgery, 224(1), 489–493[10.1016/j.amjsurg.2022.01.009]
  23. 23.Rashedur Rahman, Naomi Yagi, Keigo Hayashi, Akihiro Maruo, Hirotsugu Muratsu, Syoji Kobashi. (2024). Enhancing fracture diagnosis in pelvic X-rays by deep convolutional neural network with synthesized images from 3D-CT. Scientific Reports, 14(1), 8004[10.1038/s41598-024-58810-4]
  24. 24.Jung Min Lee, Jun Young Park, Young Jae Kim, Kwang Gi Kim. (2024). Deep-learning-based pelvic automatic segmentation in pelvic fractures. Scientific Reports, 14(1), 12258[10.1038/s41598-024-63093-w]
  25. 25.Sefatul Wasi, Saadia Binte Alam, Rashedur Rahman, M Ashraful Amin, Syoji Kobashi. (2023). Kidney Tumor Recognition from Abdominal CT Images using Transfer Learning. , 54–58[10.1109/ismvl57333.2023.00021]
  26. 26.Seung Hwan Lee, Jisu Jeon, Gil Jae Lee, Jun Young Park, Young Jae Kim, Kwang Gi Kim. (2024). Automated Association for Osteosynthesis Foundation and Orthopedic Trauma Association classification of pelvic fractures on pelvic radiographs using deep learning. Scientific Reports, 14(1), 20548[10.1038/s41598-024-71654-2]
  27. 27.Chen, Jieneng, Lu, Yongyi, Yu, Qihang, Luo, Xiangde, Adeli, Ehsan, Wang, Yan, Lu, Le, Yuille, Alan L., Zhou, Yuyin. (2021). TransUNet: Transformers Make Strong Encoders for Medical Image Segmentation. arXiv (Cornell University)[10.48550/arxiv.2102.04306]
  28. 28.Xie, Enze, Wang, Wenhai, Yu, Zhiding, Anandkumar, Anima, Alvarez, Jose M., Luo, Ping. (2021). SegFormer: Simple and Efficient Design for Semantic Segmentation with\n Transformers. arXiv (Cornell University)[10.48550/arxiv.2105.15203]
← All publications