臺灣博碩士論文加值系統

自1980年代以後，超音波成像技術已經成為一個可靠且便利的醫學診斷工具。尤其在診斷皮下病灶時，超音波成像可以提供非侵入式的、可以肉眼直接觀察成像細節的管道。
相對於其他部位增生的良、惡性腫瘤，唾液腺腫瘤因其多樣化的病理特徵，以及極其稀少的發生率，使得診斷相對而言更加困難。在臨床實務上，超音波可以提供很豐富的病理特徵，卻也因其成像的複雜度，使其難以成為主要的診斷依據。研究者試圖依據超音波影像的特性，建立電腦輔助診斷系統，然而在樣本較少的研究項目，例如唾液腺腫瘤，目前的成果仍然有改進的空間。
本篇論文將以深度學習架構為基礎，取其在影像處理領域已屢次證明的高穩定性優點，嘗試建立一套輔助診斷系統，提供腫瘤的定位，以及其良惡性的判別。根據本文實驗，我們得到了一個兩階段的系統，腫瘤的位置預測與良惡性分析之外部測試集，並分別達到80%的IoU測試，以及76.5%(92%惡性腫瘤招回率)的良惡性判斷結果。

Ultrasonography has been a reliable and efficient diagnostic technique since 1980’s. Ultrasound imaging gives a non-invasive way to visually observe the patterns of the tissues under skin. The salivary gland tumor, which is one relatively rare type of cancer and relatively varied outcome, raising the difficulty of diagnosis. In the clinical practice, the ultrasound image can indeed give abundant information, but not a decisive source for the diagnosis. Scientists attempted to construct a computer-aided diagnosis based on the ultrasonography, while the accuracy still improvable in rare cases such as salivary gland tumor (SGT) diagnosis. In this thesis, we would like to introduce the deep convolutional neural networks, which have been a robust structure in image processing, to construct a system to detect the position of the tumor and classify the character of it. We finally performed a two-staged system with almost 80% correctness in the IoU test, and 76.5% accuracy (92% recall on malignances) on the diagnosis.

書名頁 i
審定書 ii
研究許可書 iii
摘 要 v
Abstract vii
致謝 ix
Content x
List of Tables xii
List of Figures xiii
Chapter 1 Introduction 1
1.1 Salivary gland tumor 1
1.2 Ultrasound Imaging 2
1.3 Clinical sights 3
Chapter 2 Related Works 5
2.1 Traditional CAD systems 5
2.2 Convolutional Neural Networks 9
2.3 Deep Neural Networks 12
Chapter 3 System Construction 15
3.1 Data Collection 15
3.2 Data Augmentation 16
3.3 Tumor Segmentation 18
3.4 Diagnosis 27
3.5 Overall Structure 34
Chapter 4 Experiments 35
4.1 Environment 35
4.2 Criteria 35
4.3 Optimizer 37
4.4 Results 38
Chapter 5 Conclusion 44
Reference 46
Appendix A WHO Classification of Salivary Gland Tumors 54
Appendix B The Texture Features from Gray-Level Co-Occurrence Matrix 56
Appendix C Dataset 59

1. C. D. M. Fletcher, J. A. Bridge, P. C. W. Hogendoorn, & F. Mertens, “WHO Classification of Tumors of Soft Tissue and Bone,” Lyon: IARC Press, pp. 321-324, 2013.
2. J. Araya, R. Martinez, S. Niklander, M. Marshall, & A. Esguep, “Incidence and Prevalence of Salivary Gland Tumours in Valparaiso, Chile,” Medicina Oral, Patologia Oral y Cirugia Bucal, Vol. 20, No. 5, e532-e539, 2015.
3. L. Jurine, “Extrait des Expériences de Jurine sur les Chauves-Souris qu'on a Privé de la vue”, Journal de Physique, Vol. 46, pp. 145-148, 1798.
4. K. T. Dussik, “Über die Möglichkeit, Hochfrequente Mechanische Schwingungen als Diagnostisches Hilfsmitel zu Verwerten,” Zeitshrift für die Gesamte Neurologie und Psychiatrie, Vol. 174, No. 1, pp. 153-168, 1942.
5. W. D. Keidel, “Über eine neue Methode zur Registrierungger Volum Veranderungen des Herzens am Menschen,” Zeitschr. F. Kreslaufforschung, Vol. 39, pp257, 1950.
6. I. Edler & C. H. Hertz, "Use of Ultrasonic Reflectoscope for Continuous Recording of Movements of Heart Walls," K. F. S. L. Forhandlingar, Vol. 24, pp. 1-19, 1954.
7. R. Gramiak & P. M. Shah, “Echocardiography of the Aortic Root,” Investigative Radiology, Vol. 3, No. 5, pp. 356-366, 1968.
8. E. J. Białek, W. Jakubowski, & G. Karpińska, “Rule of Ultrasonography in Diagnosis and Differentiation of Pleomorphic Adenomas Work in Progress,” Head & Neck Surgery, Vol. 129, No. 3, pp. 923-933, 2003.
9. P. Katz, D. M. Hartl, & A. Guerre, “Clinical Ultrasound of the Salivary Glands,” Otolaryngologic Clinics of North America, Vol. 42, No. 6, pp. 973-1000, 2009.
10. C. J. R. Stewart, K. MacKenzie, G. W. McGarry, & A. Mowat, “Fine‐needle Aspiration Cytology of Salivary Gland: a Review of 341 Cases,” Diagnostic Cytopathology, Vol. 22, No. 3, pp. 139-146, 2000.
11. Q. Huang, F. Zhang, & X. Li, “Machine Learning in Ultrasound Computer-Aided Diagnostic Systems: a Survey,” BioMed Research International, Vol. 2018, 2018.
12. S. Joo, Y. S. Yang, W. K. Moon, and H. C. Kim, “Computer Aided Diagnosis of Solid Breast Nodules: Use of an Artificial Neural Network Based on Multiple Sonographic Features,” IEEE Transactions on Medical Imaging, Vol. 23, No. 10, pp. 1292–1300, 2004.
13. K. Horsch, M. L. Giger, L. A. Venta, and C. J. Vyborny, “Computerized Diagnosis of Breast Lesions on Ultrasound,” Medical Physics, Vol. 29, No. 2, pp. 157–164, 2002.
14. C. M. Chen, Y. H. Chou, K. C. Han, G. S. Hung, C. M. Tiu, H. J. Chiou, & S. Y. Chiou, “Breast Lesions on Sonograms: Computer-aided Diagnosis with Nearly Setting Independent Features and Artificial Neural Networks,” Radiology, Vol. 226, No. 2, pp. 504–514, 2003.
15. P. M. Shankar, V. A. Dumane, J. M. Reid , V. Genis, F. Forsberg, C.W. Piccoli, & 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.
16. B. Sahiner, H. P. Chan, M. A. Roubidoux, L. M. Hadjiiski, M. A. Helvie, C. Paramagul, J. Bailey, A. V. Nees, & C. Blane, “Malignant and Benign Breast Masses on 3D US Volumetric Images: Effect of Computer-Aided Diagnosis on Radiologist Accuracy,” Radiology, Vol. 242, No. 3, pp. 716–724, 2007.
17. M. Costantini, P. Belli, R. Lombardi, G. Franceschini, A. Mul`e, & L. Bonomo, “Characterization of Solid Breast Masses: Use of the Sonographic Breast Imaging Reporting and Data System Lexicon,” Journal of Ultrasound in Medicine, Vol. 25, No. 5, pp. 649–659, 2006.
18. R. M. Haralick & K. Shanmugam, “Textural Features for Image Classification,” IEEE Transactions on Systems, Man, and Cybernetics, Vol. 6, pp. 610-621, 1973.
19. W. Gómez, W. C. A. Pereira, & A. F. C. Infantosi, “Analysis of Co-Occurrence Texture Statistics as a Function of Gray-Level Quantization for Classifying Breast Ultrasound,” IEEE Transactions on Medical Imaging, Vol. 31, No. 10, pp. 1889-1899, 2012.
20. W. Gomez, L. Leija, A. V. Alvarenga, A. F. C. Infantosi, & W. C. A. Pereira, “Computerized Lesion Segmentation of Breast Ultrasound based on Marker‐Controlled Watershed Transformation,” Medical Physics, Vol. 37, No. 1, pp. 82-95, 2010.
21. A. Jemal, F. Bray, M. M. Center, J. Ferlay, E. Ward, & D. Forman, “Global Cancer Statisctics”, CA: A Journal for Clinicians, Vol. 61, No. 2, pp. 69-90, 2011.
22. H. Lodish, A. Berk, S. L. Zipursky, P. Matsudaira, D. Baltimore, J. E. Darnell, “Overview of Neuron Structure and Function”, in Molecular Cell Biology (Ed. 4), Section 21.1, New York, W. H. Freeman, 2000. Retrieved from: https://www.ncbi.nlm.nih.gov/books/NBK21535/ at 7/4, 2019.
23. D. H. Hubel and T. N. Wiesel, “Receptive Fields, Binocular Interaction and Functional Architecture in the Cat's Visual Cortex”, Journal of Physiology, Vol. 160, No. 1, pp. 106-154, 1962.
24. K. Fukushima, “Neocognitron: A Self-organizing Neural Network Model for a Mechanism of Pattern Recognition Unaffected by Shift in Position”, Biological Cybernetics, Vol. 36, No. 4, pp. 193-202, 1980.
25. Y. LeCun, L. Bottou, Y. Bengio, & P. Haffner, “Gradient-based Learning Applied to Document Recognition,” Proceedings of the IEEE, Vol. 86, No. 11,pp. 2278-2324, 1998.
26. V. Nair & G. E. Hinton, (2010), “Rectified Linear Units Improve Restricted Boltzmann Machines”, Proceedings of the 27th international conference on machine learning (ICML-10), pp. 807-814, 2010.
27. X. Glorot, A. Bordes, & Y. Bengio, “Deep Sparse Rectifier Neural Networks,” Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, pp. 315-323, June, 2011.
28. G. E. Hinton, N. Srivastava, A. Krizhevsky, I. Sutskever, & R. R. Salakhutdinov, “Improving Neural Networks by Preventing Co-adaptation of Feature Detectors”, arXiv preprint arXiv:1207.0580, 2012.
29. R. Zhao, W. Ouyang, H. Li, & X. Wang, “Saliency Detection by Multi-context Deep Learning,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 1265-1274, 2015.
30. A. A. Cruz-Roa, J. E. A. Ovalle, A. Madabhushi, F. A. G. Osorio, “A Deep Learning Architecture for Image Representation, Visual Interpretability and Automated Basal-cell Carcinoma Cancer Detection,” International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), pp. 403-410, 2013.
31. https://www.umiultrasound.com/ultrasound-machine/by-manufacturer/toshiba-canon-ultrasound/toshiba-canon-aplio-500/, Introduction of Toshiba Aplio-500. Last accessed at 7/4, 2019.
32. A. Dutta, A. Gupta, & A. Zisserman, “The VIA Annotation Software for Images, Audio and Video”, arXiv preprint arXiv:1904.10699, 2019. Last accessed at http://www.robots.ox.ac.uk/~vgg/software/via/, 7/15/2019.
33. O. Ronneberger, P. Fischer, & T. Brox, “U-Net: Convolutional Networks for Biomedical Image Segmentation,” International Conference on Medical Image Computing and Computer-assisted Intervention, Springer, Cham., pp. 234-241, 2015.
34. G. Huang, Z. Liu, L. V. D. Maaten, & K. Q. Weinberger, “Densely Connected Convolutional Networks,” Proceedings of the IEEE conference on Computer Vision and Pattern Recognition, pp. 4700-4708, 2017.
35. K. He, X. Zhang, S. Ren, & J. Sun, “Deep Residual Learning for Image Recognition,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 770-778, 2016.
36. D. P. Kingma & J. Ba, “Adam: A Method for Stochastic Optimization,” 3rd Internal Conference of Learning Representation, San Diego, 2015.