生活中人們常因嚴重外傷，如車禍、高處跌落，或患有骨質疏鬆症合併輕微外傷，如跌倒等的影響，造成脊椎的骨折。其中又以「壓迫性骨折」與「爆裂性骨折」較為常見，兩者之間的分辨是很重要的議題。現今臨床分析方式多是先藉由Ｘ光影像進行初步診斷，若要確立診斷就需要輔以電腦斷層掃描(CT)或核磁共振成像(MRI)的影像來確認。在臨床的實用性上，開發X 光影像的人工智慧判讀模組對於醫師的輔助效益較大，可以降低醫療成本和協助醫師診斷。

本研究使用國立成功大學醫學院附設醫院之骨科歷史病人的Ｘ光影像進行研究，並結合Yolo 模型與ResUNet 模型從X 光影像中精準分割脊椎節，從中取出前、中、後高度與高度比值、鄰近節高度比值等數據，再以Random Forest 分析數值，判斷其為正常、壓迫性骨折或爆裂性骨折。

本研究以實驗證明，與資料同為使用X 光影像之研究相比，分割與骨折偵測方面都有比較好的結果；與資料使用CT 或MRI 影像之研究相比，骨折偵測方面也表現更佳。期望能以此結果協助脊椎骨折臨床精準診斷及判讀，並且輔助急診室醫師的臨床決策，進而提升醫療品質。

People often suffer from severe trauma, such as a car accident, falling or suffering from osteoporosis which causes fractures of the spine. Besides, compression fracture and burst fracture are more common than others. The distinction between the two is a very important issue. Nowadays, most clinical analysis methods use X-ray images to make a preliminary diagnosis. If the diagnosis is to be established, it needs to be confirmed by computer tomography or magnetic resonance imaging. In terms of clinical practicability, the development of artificial intelligence models for X-ray images has a greater auxiliary benefit for physicians, which can reduce medical costs and assist physicians in diagnosis.

This study uses X-ray images of orthopedic patients from National Cheng Kung University Hospital (NCKUH) as a dataset and uses the Yolo model and the ResUNet model to accurately segment the vertebral bodies from the X-ray images. Then, we extract the features such as the anterior, middle, posterior height, height ratios and the height ratio of the adjacent vertebral body from the segmented images. The model analyzes these features with Random Forest to determine whether the vertebral body is normal, compression fracture or burst fracture.

We present the result by experiment that the result of segmentation and fracture detection is better compared with the research using X-ray images. Compared with the studies using CT or MRI images, our fracture detection also performs better. Better. We hoped that this study can assist in accurate clinical diagnosis, interpretation of spine fractures and emergency room physicians in clinical decision-making thereby improving the quality of medical care.

[1] Aebi, M. Classification of thoracolumbar fractures and dislocations. European spine journal 19, 1 (2010), 2-7.
[2] Bochkovskiy, A., Wang, C.-Y., and Liao, H.-Y. M. Yolov4: Optimal speed and accuracy of object detection. arXiv preprint arXiv:2004.10934 (2020).
[3] Denis, F. The three column spine and its significance in the classification of acute thoracolumbar spinal injuries. spine 8, 8 (1983), 817-831.
[4] Diakogiannis, F. I., Waldner, F., Caccetta, P., and Wu, C. Resunet-a: A deep learning framework for semantic segmentation of remotely sensed data. ISPRS Journal of Photogrammetry and Remote Sensing 162 (2020), 94-114.
[5] Dutta, A., and Zisserman, A. The via annotation software for images, audio and video. In Proceedings of the 27th ACM international conference on multimedia (2019), pp. 2276-2279.
[6] He, K., Zhang, X., Ren, S., and Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition (2016), pp. 770-778.
[7] Ker, J., Wang, L., Rao, J., and Lim, T. Deep learning applications in medical image analysis. IEEE Access 6 (2017), 9375-9389.
[8] Kim, K. C., Cho, H. C., Jang, T. J., Choi, J. M., and Seo, J. K. Automatic detection and segmentation of lumbar vertebrae from x-ray images for compression fracture evaluation. Computer Methods and Programs in Biomedicine 200 (2021), 105833.
[9] Kohavi, R., et al. A study of cross-validation and bootstrap for accuracy estimation and model selection. In Ijcai (1995), vol. 14, Montreal, Canada, pp. 1137-1145.
[10] Li, T., Wei, B., Cong, J., Li, X., and Li, S. S 3 eganet: 3d spinal structures segmentation via adversarial nets. IEEE Access 8 (2019), 1892-1901.
[11] Litjens, G., Kooi, T., Bejnordi, B. E., Setio, A. A. A., Ciompi, F., Ghafoorian, M., Van Der Laak, J. A., Van Ginneken, B., and Sanchez, C. I. A survey on deep learning in medical image analysis. Medical image analysis 42 (2017), 60-88.
[12] Murata, K., Endo, K., Aihara, T., Suzuki, H., Sawaji, Y., Matsuoka, Y., Nishimura, H., Takamatsu, T., Konishi, T., Maekawa, A., et al. Artificial intelligence for the detection of vertebral fractures on plain spinal radiography. Scientic Reports 10, 1 (2020), 1-8.
[13] Nicolaes, J., Raeymaeckers, S., Robben, D., Wilms, G., Vandermeulen, D., Libanati, C., and Debois, M. Detection of vertebral fractures in ct using 3d convolutional neural networks. In International Workshop and Challenge on Computational Methods and Clinical Applications for Spine Imaging (2019), Springer, pp. 3-14.
[14] Pisov, M., Kondratenko, V., Zakharov, A., Petraikin, A., Gombolevskiy, V., Morozov, S., and Belyaev, M. Keypoints localization for joint vertebra detection and fracture severity quantication. In International Conference on Medical Image Computing and Computer-Assisted Intervention (2020), Springer, pp. 723-732.
[15] Pizer, S. M., Amburn, E. P., Austin, J. D., Cromartie, R., Geselowitz, A., Greer, T., ter Haar Romeny, B., Zimmerman, J. B., and Zuiderveld, K. Adaptive histogram equalization and its variations. Computer vision, graphics, and image processing 39, 3 (1987), 355-368.
[16] Redmon, J., Divvala, S., Girshick, R., and Farhadi, A. You only look once: Unied, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition (2016), pp. 779-788.
[17] Ronneberger, O., Fischer, P., and Brox, T. U-net: Convolutional networks for biomedical image segmentation. In International Conference on Medical image computing and computer-assisted intervention (2015), Springer, pp. 234-241.
[18] Shen, D., Wu, G., and Suk, H.-I. Deep learning in medical image analysis. Annual review of biomedical engineering 19 (2017), 221-248.
[19] Yabu, A., Hoshino, M., Tabuchi, H., Takahashi, S., Masumoto, H., Akada, M., Morita, S., Maeno, T., Iwamae, M., Inose, H., et al. Using articial intelligence to diagnose fresh osteoporotic vertebral fractures on magnetic resonance images. The Spine Journal (2021).
[20] Yousefi, H., Salehi, E., Sheyjani, O. S., and Ghanaatti, H. Lumbar spine vertebral compression fracture case diagnosis using machine learning methods on ct images. In 2019 4th International Conference on Pattern Recognition and Image Analysis (IPRIA) (2019), IEEE, pp. 179-184.
[21] Zhao, H., Shi, J., Qi, X., Wang, X., and Jia, J. Pyramid scene parsing network. In Proceedings of the IEEE conference on computer vision and pattern recognition (2017), pp. 2881-2890.