Kemik Kırığı Sınıflandırmasında Derin Öğrenme Modellerinin Karşılaştırılması ve Bayes Optimizasyonu Destekli Açıklanabilir Yapay Zekâ Yaklaşımı

Abstract

İnsan vücudunda meydana gelen kemik kırıkları genellikle röntgen görüntüleri kullanılarak teşhis edilmektedir. Özellikle acil servislerde karşılaşılan kırık vakaları hızlı teşhis ve zamanında müdahale gerektirdiğinden bu sürecin otomatikleştirilmesi büyük önem taşımaktadır. Uzmanlar tarafından manuel olarak gerçekleştirilen kırık tespiti zaman alıcı bir süreçtir. Bu nedenle otomatik teşhis sistemleri erken müdahale ve sağlık hizmetlerinin etkinliğini artırma açısından kritik bir potansiyele sahiptir. Bu çalışmada röntgen görüntülerinden kemik kırıklarının sınıflandırılmasına yönelik derin öğrenme tabanlı yöntemler kullanılarak kapsamlı bir değerlendirme gerçekleştirilmiştir. Farklı veri setleri üzerinde yapılan deneysel çalışmalar sonucunda EfficientNet-B7 derin öğrenme mimarisinin kemik kırığı sınıflandırma görevinde diğer modellere göre daha yüksek performans sergilediği gözlemlenmiştir. Model performansını artırmak amacıyla olasılık temelli bir hiperparametre optimizasyon yaklaşımı olan Bayes optimizasyonu uygulanmış ve öğrenme oranı ile momentum değerleri otomatik olarak ayarlanmıştır. Optimizasyon sonrası modelin F1 performans ölçütü 0.89 değerine ulaşmıştır. Ayrıca modelin karar verme sürecini analiz edebilmek ve açıklanabilirliğini artırmak amacıyla gradyan tabanlı sınıf aktivasyon haritalama yöntemi uygulanmıştır. Elde edilen görselleştirmeler modelin kırık bölgelerine doğru biçimde odaklandığını göstermiştir. Sonuçlar derin öğrenme temelli otomatik kırık tespit sistemlerinin güvenilirliğini ve klinik uygulamalardaki potansiyelini ortaya koymaktadır.

Keywords

• Kırık kemik sınıflandırması
• derin öğrenme
• EfficientNET
• Bayes optimizasyonu
• Grad-CAM

Supporting Institution

Bu çalışma AKGÜN Bilgisayar A.Ş tarafından desteklenmiştir.

Ethical Statement

Çalışma, etik kurul izni veya herhangi bir özel izin gerektirmemektedir.

Thanks

Bu çalışmanın yürütülmesi için gerekli tüm kaynak ve finansmanı sağladıkları için AKGÜN Bilgisayar A.Ş.’ye teşekkür ederiz.

Comparative Analysis of Deep Learning Models for Bone Fracture Classification with Bayesian Optimization-Supported Explainable Artificial Intelligence

Abstract

Bone fractures are typically diagnosed using radiographic imaging techniques. In emergency departments, where rapid and accurate diagnosis is essential, automating this process has become increasingly important. Manual fracture detection performed by clinicians is time-consuming and may delay intervention. Therefore, automated diagnostic systems have significant potential to improve early decision-making and healthcare efficiency. In this study, a comprehensive evaluation was conducted using deep learning-based methods for the classification of bone fractures from X-ray images. Experimental results obtained from two publicly available datasets demonstrated that the EfficientNet-B7 deep learning architecture outperformed other convolutional neural network models in fracture classification tasks. To enhance model performance, a probabilistic hyperparameter optimization approach based on Bayesian optimization was applied. The learning rate and momentum parameters were automatically tuned to maximize validation performance. After optimization, the model achieved an F1 performance score of 0.89 on the test data. Furthermore, to improve interpretability and analyze the decision-making process of the model, a gradient-based class activation mapping technique was employed. The visualization results showed that the model accurately focused on clinically relevant fracture regions during classification. The findings indicate that the proposed deep learning framework provides reliable performance and improved interpretability, highlighting its potential use as a supportive decision tool in clinical practice, particularly in emergency settings.

Keywords

• Bone fracture classification
• deep learning
• EfficientNet
• Bayesian optimization
• Grad-CAM

References

1. Lindsey, R., Daluiski, A., Chopra, S., Lachapelle, A., Mozer, M., Sicular, S., Hanel, D., Gardner, M., Gupta, A., Hotchkiss, R., & Potter, H. (2018). Deep neural network improves fracture detection by clinicians. Proceedings of the National Academy of Sciences, 115(45), 11591–11596. https://doi.org/10.1073/pnas.1806905115
2. Gao, Y., Soh, N. Y. T., Liu, N., Lim, G., Ting, D., Cheng, L. T. E., Wong, K. M., Liew, C., Oh, H. C., Tan, J. R., Venkataraman, N., Goh, S. H., & Yan, Y. Y. (2023). Application of a deep learning algorithm in the detection of hip fractures. iScience, 26(8), 107350. https://doi.org/10.1016/j.isci.2023.107350
3. Su, Z., Adam, A., Nasrudin, M. F., Ayob, M., & Punganan, G. (2023). Skeletal fracture detection with deep learning: A comprehensive review. Diagnostics, 13(20), 3245. https://doi.org/10.3390/diagnostics13203245
4. Bingöl, E., Demirel, S., Urfalı, A., & Bozkır, Ö. F. (2024). Analyzing the performance of convolutional neural networks and transformer models in automated bone fracture detection. Mus Alparslan University Journal of Science, 12(2), 64–71. https://doi.org/10.18586/msufbd.1440119
5. Kadiroğlu, Z., Deniz, E., & Şenyiğit, A. (2024). A comparison of deep learning models for pneumonia detection from chest X-ray images. Journal of the Faculty of Engineering and Architecture of Gazi University, 39(2), 729–740.
6. Bhati, D., Neha, F., & Amiruzzaman, M. (2024). A survey on explainable artificial intelligence (XAI) techniques for visualizing deep learning models in medical imaging. Journal of Imaging, 10(10), 239. https://doi.org/10.3390/jimaging10100239
7. Houssein, E. H., Gamal, A. M., Younis, E. M., & Mohamed, E. (2025). Explainable artificial intelligence for medical imaging systems using deep learning: A comprehensive review. Cluster Computing, 28(7), 469.
8. İli, M. D., & Özyurt, F. (2025). Açıklanabilir yapay zeka yöntemleriyle MR görüntülerinden beyin tümörü tespiti. Kahramanmaraş Sütçü İmam Üniversitesi Mühendislik Bilimleri Dergisi, 28(2), 1092–1109.

9. Akın, K. D., Gurkan, C., Budak, A., & Karataş, H. (2022). Classification of monkeypox skin lesion using the explainable artificial intelligence assisted convolutional neural networks. Avrupa Bilim ve Teknoloji Dergisi, 40, 106–110.
10. Sahin, E., Celikten, A., Demirel, S., Akpulat, A., Budak, K., & Karatas, H. (2024). Pneumonia detection in chest X-ray images using deep learning and Grad-CAM++ visualization. In 2024 Innovations in Intelligent Systems and Applications Conference (ASYU) (pp. 1–5). IEEE.
11. Tesfaw, A., Teshager, A., Mirolgn, A., & Ambachew, E. (2022). Bone fracture classification using heterogeneous ensemble machine learning and deep learning algorithms. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.4700025
12. Chung, S. W., Han, S. S., Lee, J. W., Oh, K. S., Kim, N. R., Yoon, J. P., ... & Kim, Y. (2018). Automated detection and classification of the proximal humerus fracture by using deep learning algorithm. Acta Orthopaedica, 89(4), 468–473. https://doi.org/10.1080/17453674.2018.1453714
13. Yao, L., Guan, X., Song, X., Tan, Y., Wang, C., Jin, C., Chen, M., Wang, H., & Zhang, M. (2021). Rib fracture detection system based on deep learning. Scientific Reports, 11(1), 23513. https://doi.org/10.1038/s41598-021-03002-7
14. Yadav, D. P., & Rathor, S. (2020). Bone fracture detection and classification using deep learning approach. In 2020 International Conference on Power Electronics & IoT Applications in Renewable Energy and Its Control (PARC) (pp. 282–285). IEEE.
15. Karanam, S. R., Srinivas, Y., & Chakravarty, S. (2023). A supervised approach to musculoskeletal imaging fracture detection and classification using deep learning algorithms. Computer Assisted Methods in Engineering and Science, 30(3), 369–385.
16. Sato, Y., Takegami, Y., Asamoto, T., Ono, Y., Hidetoshi, T., Goto, R., Kitamura, A., & Honda, S. (2021). Artificial intelligence improves the accuracy of residents in the diagnosis of hip fractures: A multicenter study. BMC Musculoskeletal Disorders, 22(1), 407. https://doi.org/10.1186/s12891-021-04260-2
17. Saii, M., & Mayya, M. (2018). Hand and finger fracture detection and type classification using image processing. International Journal of Computer Science Trends and Technology, 6(4), 68–76.
18. De Cid Rodríguez, J. (2021). Fracture detection from X-rays with deep learning [Master’s thesis, Universitat Politècnica de Catalunya].
19. Singh, A., Ardakani, A. A., Loh, H. W., Anamika, P. V., Acharya, U. R., Kamath, S., & Bhat, A. K. (2023). Automated detection of scaphoid fractures using deep neural networks in radiographs. Engineering Applications of Artificial Intelligence, 122, 106165. https://doi.org/10.1016/j.engappai.2023.106165
20. Öztürk, Ö., & Kutucu, H. (2017). Detection of bone fractures using image processing techniques and artificial neural networks. In 2017 International Artificial Intelligence and Data Processing Symposium (IDAP) (pp. 1–5). IEEE.
21. Rajpurkar, P., Irvin, J., Bagul, A., et al. (2018). MURA: Large dataset for abnormality detection in musculoskeletal radiographs. In Proceedings of the 1st Conference on Medical Imaging with Deep Learning (MIDL 2018), Amsterdam, The Netherlands.
22. Roboflow. (2023, February 1). Bone fracture dataset (Roboflow Universe). Roboflow Universe. https://universe.roboflow.com/makerere-university/fracture-qc49p
23. Doğan, F., Aktaş, M., & Gürsoy, M. İ. (2024). Classification of skin diseases with different deep learning models and comparison of the performances of the models. Türk Doğa ve Fen Dergisi, 13(3), 117–123.
24. Tan, M., & Le, Q. V. (2019). EfficientNet: Rethinking model scaling for convolutional neural networks. In Proceedings of the 36th International Conference on Machine Learning (ICML 2019) (pp. 6105–6114).
25. Snoek, J., Larochelle, H., & Adams, R. P. (2012). Practical Bayesian optimization of machine learning algorithms. Advances in Neural Information Processing Systems, 25.
26. Selvaraju, R. R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., & Batra, D. (2017). Grad-CAM: Visual explanations from deep networks via gradient-based localization. In Proceedings of the IEEE International Conference on Computer Vision (pp. 618–626). https://doi.org/10.1109/ICCV.2017.74