肌肉挫傷是一種常見的閉合性損傷，經常由於組織受到瞬間且強烈的壓力撞擊而產生。若能及時掌控肌肉復原情形及進行檢傷分類，才能施予有效的治療。目前臨床上診斷肌肉挫傷主要利用超音波影像，然而超音波影像的判讀有賴於操作者的經驗，因此以超音波醫學影像來定量評估挫傷的復原情況將是未來的趨勢。研究設計了一種挫傷模型，利用600g砝碼分別從高度30cm及45cm處垂直落下，造成中度及嚴重挫傷。接著對大鼠患部分別在挫傷後第1、2、3、5、7、9、11、13、15、18及21天利用30MHz高頻超音波進行掃描，並利用U-Net、UNet++及MultiResUNet三個深度學習模型自動圈選肥腸肌區域。而後計算該區域內肌肉組織之積體逆散射、Nakagami統計參數及利用灰階共生矩陣提取紋理特徵，並搭配天數來觀察大鼠的復原情形。最後利用K-means演算法並透過特徵參數區分破壞階段、修復階段及重塑階段，以觀察不同程度肌肉挫傷的病程變化。影像分割結果顯示，MultiResUNet增加了殘差連接及批次正規化，準確率達到 87.55%。特徵參數結果顯示，積體逆散射和亂度受到回聲震幅大小的影響，因此可能無法直接觀察到癒合過程。然而Nakagami統計參數、對比度、同質性、相關性和能量能識別不同挫傷嚴重度的肌肉癒合。分群結果顯示，當嚴重挫傷時，挫傷後第1天到第4天的影像聚集到破壞階段，第5天到第15天的影像聚集到修復階段，第7天至第21天的影像聚集到重塑階段。當中等挫傷時，挫傷後第1天至第 3天的影像聚集到破壞階段，第5天至第9天的影像聚集到修復階段，第11天到第21天的影像聚集到重塑階段。並且K-means 演算法可以更好地描述癒合階段的接續性以及部分重疊性。

Muscle contusion is a common closed injury in which frequently occurs in tissues impacted by a rapid and strong compressive force. The understanding of muscle healing and accurate triage are necessary for the valid treatment. Nowadays, the diagnosis of muscle contusion mainly uses ultrasound imaging. Nevertheless, the diagnosis of ultrasound images depends on the experienced operators. Therefore, quantitatively assessing the healing with ultrasound images is needed. In this study, using 600g weight dropped from height of 30cm and 45cm were able to result in moderate and severe contusions. Ultrasound images were used to observe the injured region of the rat during three weeks after contusion. Deep learning methods of U-Net, UNet++, and MultiResUNet, were used to automatically annotate the gastrocnemius. And then, calculating the corresponding integrated backscatter, Nakagami-m, and texture features extracted from the gray-level co-occurrence matrix. Lastly, the healing was divided into three phases by K-means clustering, including destruction, repair, and remodeling. The segmented results of MultiResUNet reached 87.55% accuracy. The featured results indicated that the integrated backscatter and entropy are sensitive to the scaling of echo amplitudes. In comparison, Nakagami-m, contrast, homogeneity, correlation, and energy were able to distinguish contusion healing in different severities. The clustering results indicated that on severe contusion, images from Day 1 to Day 4 after contusion were clustered to destruction, that from Day 5 to Day 15 were clustered to repair, and that from Day 7 to Day 21 were clustered to remodeling. On moderate contusion, there were images from Day 1 to Day 3 were clustered to destruction, that from Day 5 to Day 9 were clustered to repair, and that from Day 11 to Day 21 were clustered to remodeling. K-means clustering appropriately described the subsequent overlapping phases of healing process.

[1] T. A. Järvinen, T. L. Järvinen, M. Kääriäinen, V. Äärimaa, S. Vaittinen, H. Kalimo, and M. Järvinen, "Muscle injuries: optimising recovery," Best Practice & Research Clinical Rheumatology, vol. 21, no. 2, pp. 317-331, 2007.
[2] Z. Y. Yu, "Classification of Muscle Contusion Healing Stages using Ultrasound Statistical Parameters and Image Texture Features Extraction," 2018.
[3] T. A. Järvinen, M. Kääriäinen, M. Järvinen, and H. Kalimo, "Muscle strain injuries," Current opinion in rheumatology, vol. 12, no. 2, pp. 155-161, 2000.
[4] J. B. Woodhouse and E. G. McNally, "Ultrasound of skeletal muscle injury: an update," in Seminars in Ultrasound, CT and MRI, 2011, vol. 32, no. 2: Elsevier, pp. 91-100.
[5] J. J. Crisco, P. Jokl, G. T. Heinen, M. D. Connell, and M. M. Panjabi, "A muscle contusion injury model: biomechanics, physiology, and histology," The American journal of sports medicine, vol. 22, no. 5, pp. 702-710, 1994.
[6] J. M. Beiner and P. Jokl, "Muscle contusion injuries: current treatment options," JAAOS-Journal of the American Academy of Orthopaedic Surgeons, vol. 9, no. 4, pp. 227-237, 2001.
[7] J. de Souza and C. Gottfried, "Muscle injury: review of experimental models," Journal of Electromyography and Kinesiology, vol. 23, no. 6, pp. 1253-1260, 2013.
[8] T. A. Järvinen, T. L. Järvinen, M. Kääriäinen, H. Kalimo, and M. Järvinen, "Muscle injuries: biology and treatment," The American journal of sports medicine, vol. 33, no. 5, pp. 745-764, 2005.
[9] T. Watson, "Soft Tissue Repair and Healing Review," ed, 2016.
[10] L. I. Filippin, A. J. Moreira, N. P. Marroni, and R. M. Xavier, "Nitric oxide and repair of skeletal muscle injury," Nitric oxide, vol. 21, no. 3-4, pp. 157-163, 2009.
[11] S. Liu, Y. Wang, X. Yang, B. Lei, L. Liu, S. X. Li, D. Ni, and T. Wang, "Deep learning in medical ultrasound analysis: a review," Engineering, vol. 5, no. 2, pp. 261-275, 2019.
[12] Meire, H. B, Farrant, and Pat, Basic Ultrasound. 1995.
[13] S. Su, Z. Gao, H. Zhang, Q. Lin, W. K. Hau, and S. Li, "Detection of lumen and media-adventitia borders in IVUS images using sparse auto-encoder neural network," in 2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017), 2017: IEEE, pp. 1120-1124.
[14] E. Smistad and L. Løvstakken, "Vessel detection in ultrasound images using deep convolutional neural networks," in Deep Learning and Data Labeling for Medical Applications: Springer, 2016, pp. 30-38.
[15] U. M. Reddy, R. A. Filly, and J. A. Copel, "Prenatal imaging: ultrasonography and magnetic resonance imaging," Obstetrics and gynecology, vol. 112, no. 1, p. 145, 2008.
[16] Z. Wang, Z. Zhang, J. Zheng, B. Huang, I. Voiculescu, and G.-Z. Yang, "Deep learning in medical ultrasound image segmentation: A review," arXiv preprint arXiv:2002.07703, 2020.
[17] X. Liu, J. Shi, and Q. Zhang, "Tumor classification by deep polynomial network and multiple kernel learning on small ultrasound image dataset," in International Workshop on Machine Learning in Medical Imaging, 2015: Springer, pp. 313-320.
[18] J. Shi, S. Zhou, X. Liu, Q. Zhang, M. Lu, and T. Wang, "Stacked deep polynomial network based representation learning for tumor classification with small ultrasound image dataset," Neurocomputing, vol. 194, pp. 87-94, 2016.
[19] J. Ferlay, H. R. Shin, F. Bray, D. Forman, C. Mathers, and D. M. Parkin, "Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008," International journal of cancer, vol. 127, no. 12, pp. 2893-2917, 2010.
[20] L. Guo, D. Wang, H. Xu, Y. Qian, C. Wang, X. Zheng, Q. Zhang, and J. Shi, "CEUS-based classification of liver tumors with deep canonical correlation analysis and multi-kernel learning," in 2017 39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2017: IEEE, pp. 1748-1751.
[21] M. H. Yap, G. Pons, J. Marti, S. Ganau, M. Sentis, R. Zwiggelaar, A. K. Davison, and R. Marti, "Automated breast ultrasound lesions detection using convolutional neural networks," IEEE journal of biomedical and health informatics, vol. 22, no. 4, pp. 1218-1226, 2017.
[22] N. Singhal, S. Mukherjee, and C. Perrey, "Automated assessment of endometrium from transvaginal ultrasound using deep learned snake," in 2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017), 2017: IEEE, pp. 283-286.
[23] X. Yang, L. Yu, L. Wu, Y. Wang, D. Ni, J. Qin, and P.-A. Heng, "Fine-grained recurrent neural networks for automatic prostate segmentation in ultrasound images," in Proceedings of the AAAI Conference on Artificial Intelligence, 2017, vol. 31, no. 1.
[24] J. Ma, F. Wu, T. a. Jiang, Q. Zhao, and D. Kong, "Ultrasound image-based thyroid nodule automatic segmentation using convolutional neural networks," International journal of computer assisted radiology and surgery, vol. 12, no. 11, pp. 1895-1910, 2017.
[25] Y. Li, R. Xu, J. Ohya, and H. Iwata, "Automatic fetal body and amniotic fluid segmentation from fetal ultrasound images by encoder-decoder network with inner layers," in 2017 39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2017: IEEE, pp. 1485-1488.
[26] J. P. Jones and S. Leeman, "Ultrasonic tissue characterization and quantitative ultrasound scatter imaging: methods and approaches," in Physics and Engineering in Medical Imaging, 1982, vol. 372: International Society for Optics and Photonics, pp. 247-258.
[27] C.-C. Huang, S.-H. Wang, and P.-H. Tsui, "Detection of blood coagulation and clot formation using quantitative ultrasonic parameters," Ultrasound in medicine & biology, vol. 31, no. 11, pp. 1567-1573, 2005.
[28] Y.-P. Huang, Y.-P. Zheng, S.-F. Leung, and A. Choi, "High frequency ultrasound assessment of skin fibrosis: clinical results," Ultrasound in medicine & biology, vol. 33, no. 8, pp. 1191-1198, 2007.
[29] B. Hete and K. K. Shung, "A study of the relationship between mechanical and ultrasonic properties of dystrophic and normal skeletal muscle," Ultrasound in medicine & biology, vol. 21, no. 3, pp. 343-352, 1995.
[30] K. K. Shung, M. Smith, and B. M. Tsui, Principles of medical imaging. Academic Press, 2012.
[31] P.-H. Tsui and S.-H. Wang, "The effect of transducer characteristics on the estimation of Nakagami paramater as a function of scatterer concentration," Ultrasound in medicine & biology, vol. 30, no. 10, pp. 1345-1353, 2004.
[32] R. F. Wagner, M. F. Insana, and D. G. Brown, "Statistical properties of radio-frequency and envelope-detected signals with applications to medical ultrasound," JOSA A, vol. 4, no. 5, pp. 910-922, 1987.
[33] P. M. Shankar, "A general statistical model for ultrasonic backscattering from tissues," IEEE transactions on ultrasonics, ferroelectrics, and frequency control, vol. 47, no. 3, pp. 727-736, 2000.
[34] P. M. Shankar, V. Dumane, J. M. Reid, V. Genis, F. Forsberg, C. W. Piccoli, and B. B. Goldberg, "Classification of ultrasonic B-mode images of breast masses using Nakagami distribution," IEEE transactions on ultrasonics, ferroelectrics, and frequency control, vol. 48, no. 2, pp. 569-580, 2001.
[35] Y.-H. Lin, C.-C. Huang, and S.-H. Wang, "Quantitative assessments of burn degree by high-frequency ultrasonic backscattering and statistical model," Physics in Medicine & Biology, vol. 56, no. 3, p. 757, 2011.
[36] S.-H. Wang, F.-C. Tsai, and Y.-L. Hung, "Characterization of bony tissues from ultrasonic backscattering using statistical models," in 2001 IEEE Ultrasonics Symposium. Proceedings. An International Symposium (Cat. No. 01CH37263), 2001, vol. 2: IEEE, pp. 1205-1208.
[37] R. M. Haralick, K. Shanmugam, and I. H. Dinstein, "Textural features for image classification," IEEE Transactions on systems, man, and cybernetics, no. 6, pp. 610-621, 1973.
[38] R. P. Kruger, W. B. Thompson, and A. F. Turner, "Computer diagnosis of pneumoconiosis," IEEE Transactions on Systems, Man, and Cybernetics, no. 1, pp. 40-49, 1974.
[39] M. Leineweber and Y. Gao, "Quantifying skeletal muscle recovery in a rat injury model using ultrasound imaging," Journal of biomechanics, vol. 48, no. 2, pp. 379-382, 2015.
[40] M. G. B. Dantas, C. M. D. Damasceno, V. R. P. d. Barros, E. S. Menezes, H. d. S. Fontoura, R. S. d. Lima, F. O. Carvalho, and J. R. G. d. S. Almeida, "Creation of a contusion injury method for skeletal muscle in rats with differing impacts," Acta cirurgica brasileira, vol. 32, no. 5, pp. 369-375, 2017.
[41] F. M. Valckx and J. M. Thijssen, "Characterization of echographic image texture by cooccurrence matrix parameters," Ultrasound in medicine & biology, vol. 23, no. 4, pp. 559-571, 1997.
[42] R.-F. Chang, C.-C. Lee, and C.-M. Lo, "Computer-aided diagnosis of different rotator cuff lesions using shoulder musculoskeletal ultrasound," Ultrasound in medicine & biology, vol. 42, no. 9, pp. 2315-2322, 2016.
[43] R. Suganya and S. Rajaram, "Feature extraction and classification of ultrasound liver images using haralick texture-primitive features: Application of SVM classifier," in 2013 international conference on recent trends in information technology (ICRTIT), 2013: IEEE, pp. 596-602.
[44] O. Ronneberger, P. Fischer, and T. Brox, "U-net: Convolutional networks for biomedical image segmentation," in International Conference on Medical image computing and computer-assisted intervention, 2015: Springer, pp. 234-241.
[45] Z. Zhou, M. M. R. Siddiquee, N. Tajbakhsh, and J. Liang, "Unet++: A nested u-net architecture for medical image segmentation," in Deep learning in medical image analysis and multimodal learning for clinical decision support: Springer, 2018, pp. 3-11.
[46] Z. Zhou, M. M. R. Siddiquee, N. Tajbakhsh, and J. Liang, "Unet++: Redesigning skip connections to exploit multiscale features in image segmentation," IEEE transactions on medical imaging, vol. 39, no. 6, pp. 1856-1867, 2019.
[47] N. Ibtehaz and M. S. Rahman, "MultiResUNet: Rethinking the U-Net architecture for multimodal biomedical image segmentation," Neural Networks, vol. 121, pp. 74-87, 2020.
[48] K. Raum and W. D. O'Brien, "Pulse-echo field distribution measurement technique for high-frequency ultrasound sources," ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 44, no. 4, pp. 810-815, 1997.
[49] N. Otsu, "A threshold selection method from gray-level histograms," IEEE transactions on systems, man, and cybernetics, vol. 9, no. 1, pp. 62-66, 1979.
[50] J. M. Rijkelijkhuizen, H. J. Meijer, G. C. Baan, and P. A. Huijing, "Myofascial force transmission also occurs between antagonistic muscles located within opposite compartments of the rat lower hind limb," Journal of Electromyography and Kinesiology, vol. 17, no. 6, pp. 690-697, 2007.
[51] P.-H. Tsui, S.-H. Wang, C.-C. Huang, and C.-Y. Chiu, "Quantitative analysis of noise influence on the detection of scatterer concentration by Nakagami parameter," Journal of Medical and Biological Engineering, vol. 25, no. 2, pp. 45-51, 2005.
[52] K. K. Shung, Diagnostic ultrasound: Imaging and blood flow measurements. CRC press, 2015.
[53] D. Gadkari, "Image quality analysis using GLCM," 2004.
[54] A. H. Fischer, K. A. Jacobson, J. Rose, and R. Zeller, "Hematoxylin and eosin staining of tissue and cell sections," Cold spring harbor protocols, vol. 2008, no. 5, p. pdb. prot4986, 2008.
[55] C. Shackelford, G. Long, J. Wolf, C. Okerberg, and R. Herbert, "Qualitative and quantitative analysis of nonneoplastic lesions in toxicology studies," Toxicologic pathology, vol. 30, no. 1, pp. 93-96, 2002.
[56] M. Drozdzal, E. Vorontsov, G. Chartrand, S. Kadoury, and C. Pal, "The importance of skip connections in biomedical image segmentation," in Deep learning and data labeling for medical applications: Springer, 2016, pp. 179-187.
[57] H.-C. Lin and S.-H. Wang, "Window-Modulated Compounding Nakagami Parameter Ratio Approach for Assessing Muscle Perfusion with Contrast-Enhanced Ultrasound Imaging," Sensors, vol. 20, no. 12, p. 3584, 2020.
[58] M. Byra, L. Wan, J. H. Wong, J. Du, S. B. Shah, M. P. Andre, and E. Y. Chang, "Quantitative ultrasound and b-mode image texture features correlate with collagen and myelin content in human ulnar nerve fascicles," Ultrasound in medicine & biology, vol. 45, no. 7, pp. 1830-1840, 2019.
[59] D. Steilen, R. Hauser, B. Woldin, and S. Sawyer, "Chronic neck pain: making the connection between capsular ligament laxity and cervical instability," The open orthopaedics journal, vol. 8, p. 326, 2014.
[60] L. Forcina, M. Cosentino, and A. Musarò, "Mechanisms regulating muscle regeneration: Insights into the interrelated and time-dependent phases of tissue healing," Cells, vol. 9, no. 5, p. 1297, 2020.