臺灣博碩士論文加值系統

乳癌屬於較容易在早期發現徵兆的癌症，若能及早發現並治療，有助於提高術後的存活率。超音波檢測即為目前常見的乳房腫瘤檢測方法，其具有非侵入性、價格較低、檢查方便等優點，幾乎已成為醫療院所必備的診斷儀器。因此本研究提出了一套結合特徵篩選及參數設定的方法，用以縮短超音波腫瘤影像輔助診斷系統的訓練時間，並提升其分類的正確率。我們首先會對每張超音波腫瘤影像進行前處理，並利用等位函數法切割出腫瘤本體的影像，接著再分別計算其紋理特徵和形狀特徵。最後，本研究使用支援向量機來辨別腫瘤良惡性，並利用人工免疫演算法進行特徵篩選及決定支援向量機的參數。
實驗結果顯示，本研究所提出的方法能使乳房腫瘤的分類正確率達到95.24%。惡性腫瘤的敏感性指標為97.78%，良性腫瘤的敏感性指標為93.33%。惡性腫瘤的預測指標為91.67%。良性腫瘤的預測指標為98.25%。證明使用人工免疫演算法能提升支援向量機判斷乳房腫瘤良惡的正確率，並有效的縮短支援向量機的訓練時間。

Early diagnosis and treatment of breast cancer can effectively decrease the mortality rate. Recently, ultrasound examination plays an important role in the field of breast cancer diagnosis because of its non-invasive, low price, and convenience.
To promote the classification accuracy and decrease training time of an ultrasound breast tumor image computer-aided diagnosis system, we proposed an approach which combined feature selection and parameter setting simultaneously in this paper.
Before the classification, all the breast tumors were segmented automatically by a level set method. Then, the texture features and shape features were first extracted following the use of an artificial immune system algorithm to detect significant features and determine the near-optimal parameters for the support vector machine to identify the tumor as benign or malignant.
The experiment shows that the accuracy of the proposed system for classifying breast tumors is 95.24%, the sensitivity is 97.78%, the specificity is 93.33%, the positive predictive value is 91.67%, and the negative predictive value is 98.25%. It is proved that the use of an artificial immune system algorithm can promote the classification accuracy and decrease training time of the support vector machine.

目錄
指導教授推薦書
口試委員審定書
國家圖書館博碩士論文電子檔案上網授權書 iii
長庚大學碩博士論文著作授權書 iv
致謝 v
摘要 vi
Abstract vii
目錄 viii
圖目錄 xi
表目錄 xii
第一章 緒論 1
1.1 前言 1
1.2 研究動機與目的 2
第二章 背景與文獻探討 6
2.1 超音波乳房腫瘤影像診斷回顧 6
2.2 腫瘤影像處理 9
2.2.1 非等方性延展濾波 9
2.2.2 Stick偵測 10
2.2.3 等位函數法 12
2.3 支援向量機 15
2.3.1 最佳超平面 15
2.3.2 線性可分割 16
2.3.3 線性不可分割 18
2.3.4 核函數 19
2.3.5 支援向量機的弱點 20
2.4 主成份分析 21
2.5 免疫演算 22
2.5.1 生物免疫系統 22
2.5.2 免疫系統運作機制 24
2.5.3 人工免疫演算法 26
第三章 研究方法與流程 29
3.1 研究流程 30
3.2 超音波影像前處理 31
3.3 腫瘤特徵的擷取與計算 32
3.3.1 腫瘤形狀特徵 33
3.3.2 腫瘤紋理特徵 35
3.3.3 腫瘤特徵計算結果 36
3.4 腫瘤特徵篩選 37
3.5 腫瘤分類 38
3.6 人工免疫演算法結合支援向量機 39
3.6.1 產生初始抗體 39
3.6.2 複製與變異 40
3.6.3 選出具有最佳親和力的抗體 41
第四章 實驗結果與分析 42
4.1 影像資料及實驗設備 42
4.2 實驗結果 42
第五章 結論 48
參考文獻 50

圖目錄
圖1-1、乳房超音波腫瘤影像 4
圖2-1、 矩形區塊的stick線段分布 11
圖2-2、二維歐幾里得平面中的封閉曲線 14
圖2-3、等位函數法操作示意圖 14
圖2-4、最佳超平面 16
圖3-1、複製選擇演算法基本流程 29
圖3-2、研究流程圖 30
圖3-3、超音波影像前處理流程圖 31
圖3-4、影像前處理結果 32
圖3-5、腫瘤的最大和最小直徑 34
圖3-6、包覆腫瘤的最小凸多邊形 34
圖3-7、包覆腫瘤的最小矩形 35
圖3-8、人工免疫演算法結合支援向量機架構圖 38
圖3-9、二進位編碼之初始抗體示意圖 39
圖4-1、實驗結果之ROC Curve 47

表目錄
表1-1、民國93至97年台灣各期乳癌五年存活率 2
表2-1、超音波乳房腫瘤影像診斷相關文獻 7
表4-1、實驗結果及其正確率 44
表4-2、人工免疫演算法結合支援向量機的分類結果 45
表4-3、實驗結果之效能指標 46

[1] Stavros, A. T., Thickman, D., Rapp, C. L., Dennis, M. A., Parker, S. H. and Sisney, G. A., “Solid breast nodules: use of sonography to distinguish between benign and malignant lesions,” Radiology, vol. 196(1), pp. 123-134, July 1995.
[2] Chang, R. F., Wu, W. J., Moon, W. K. and Chen, D. R., “Improvement in breast tumor discrimination by support vector machines and speckle-emphasis texture analysis,” Ultrasound Med Biol, vol. 29(5), pp. 679-686, May 2003.
[3] Chen, D. R., Chang, R. F. and Huang, Y. L., “Computer-aided diagnosis applied to US of solid breast nodules by using neural networks,” Radiology, vol. 213(2), pp. 407-412, November 1999.
[4] Czerwinski, R. N., Jones, D. L. and O'Brien, W. D., Jr., “Edge detection in ultrasound speckle noise,” Image Processing, 1994. Proceedings. ICIP-94., IEEE International Conference, vol. 3, pp. 304-308, November 1994.
[5] Osher, S. and Sethian, J. A., “Fronts propagating with curvature-dependent speed: Algorithms based on Hamilton-Jacobi formulations,” Journal of Computational Physics, vol. 79(1), pp. 12-49, 1988.
[6] Ruggiero, C., Bagnoli, F., Sacile, R., Calabrese, M., Rescinito, G. and Sardanelli, F., “Automatic recognition of malignant lesions in ultrasound images by artificial neural networks,” Engineering in Medicine and Biology Society, 1998. Proceedings of the 20th Annual International Conference of the IEEE, vol. 2, pp. 872-875, October-November 1998.
[7] Horsch, K., Giger, M. L., Venta, L. A. and Vyborny, C. J., “Computerized diagnosis of breast lesions on ultrasound,” Medical Physics, vol. 29(2), pp. 157-164, 2002.
[8] Drukker, K., Giger, M. L., Vyborny, C. J. and Mendelson, E. B., “Computerized Detection and Classification of Cancer on Breast Ultrasound,” Academic Radiology, vol. 11(5), pp. 526-535, 2004.
[9] Chang, R. F., Wu, W. J., Moon, W. K. and Chen, D. R., “Automatic ultrasound segmentation and morphology based diagnosis of solid breast tumors,” Breast Cancer Research and Treatment, vol. 89(2), pp. 179-185, 2005.
[10] Kim, K. G., Cho, S. W., Min, S. J., Kim, J. H., Min, B. G. and Bae, K. T., “Computerized scheme for assessing ultrasonographic features of breast masses,” Academic Radiology, vol. 12(1), pp. 58-66, 2005.
[11] Yang, W., Zhang, S., Chen, Y. and Li, W., “Measuring shape complexity of breast lesions on ultrasound images,” vol. 6920, 2008.
[12] Gerig, G., Kubler, O., Kikinis, R. and Jolesz, F. A., “Nonlinear anisotropic filtering of MRI data,” IEEE Transactions on Medical Imaging, vol. 11(2), pp. 221-232, 1992.
[13] Gnoping, Q., “An improved recursive median filtering scheme for image processing,” IEEE Transactions on Image Processing, vol. 5(4), pp. 646-648, 1996.
[14] Perona, P. and Malik, J., “Scale-space and edge detection using anisotropic diffusion,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 12(7), pp. 629-639, 1990.
[15] Haralick, R. M., “Digital Step Edges from Zero Crossing of Second Directional Derivatives,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-6(1), pp. 58-68, 1984.
[16] Canny, J., “A Computational Approach to Edge Detection,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-8(6), pp. 679-698, 1986.
[17] Czerwinski, R. N., Jones, D. L. and O'Brien, W. D., Jr., “Detection of lines and boundaries in speckle images-application to medical ultrasound,” IEEE Transactions on Medical Imaging, vol. 18(2), pp. 126-136, 1999.
[18] Vapnik, V. N. The nature of statistical learning theory. Springer, 2000.
[19] Steinwart, I. and Christmann, A. Support vector machines. Springer, 2008.
[20] Bertsekas, D. P. and Tsitsiklis, J. N. Parallel and distributed computation : numerical methods. Prentice Hall, Englewood Cliffs, New Jersey, 1989.
[21] Pietsch, A., “Eigenvalue distribution of compact operators,” Integral Equations and Operator Theory, vol. 9(4), pp. 610-612, 1986.
[22] Qiu, B., Prinet, V., Perrier, E. and Monga, O., “Multi-block PCA method for image change detection,” Proceedings. 12th International Conference on Image Analysis and Processing, pp. 385-390, September 2003.
[23] Farmer, J. D., Packard, N. H. and Perelson, A. S., “The immune system, adaptation, and machine learning,” Physica D: Nonlinear Phenomena, vol. 22(1-3), pp. 187-204, 1986.
[24] Forrest, S., Perelson, A. S., Allen, L. and Cherukuri, R., “Self-nonself discrimination in a computer,” Proceedings of the 1994 IEEE Symposium on Research in Security and Privacy, pp. 202-212, 1994.
[25] Jerne, N. K., “The immune system,” Sci Am, vol. 229(1), pp. 52-60, July 1973.
[26] Burnet, F. M., “A modification of Jerne's theory of antibody production using the concept of clonal selection,” CA Cancer J Clin, vol. 26(2), pp. 119-121, March-April 1976.
[27] de Castro, L. N. and Von Zuben, F. J., “The Clonal Selection Algorithm with Engineering Applications,” GECCO'00, Workshop on Artificial Immune Systems and Their Applications, 2000.
[28] Yang, X., Jiao, L., Qi, Y. and Jin, H. Multifocus Image Fusion Based on Multiwavelet and Immune Clonal Selection. Springer Berlin / Heidelberg, City, 2006.
[29] Jerne, N. K., “Towards a network theory of the immune system,” Ann Immunol (Paris), vol. 125C(1-2), pp. 373-389, Janurary 1974.
[30] de Castro, L. N. and Von Zuben, F. J., “aiNet: An artificial immune network for data analysis,” Data Mining: a heuristic approach, vol. 1 pp. 231-259, 2001.
[31] Tang, Z., Yamaguchi, T., Tashima, K., Ishizuka, O. and Tanno, K., “Multiple-valued immune network model and its simulations,” Multiple-Valued Logic, 1997. Proceedings., 1997 27th International Symposium on, pp. 233-238, 1997.
[32] Chang, R.-F., Wu, W.-J., Moon, W. K., Chou, Y.-H. and Chen, D.-R., “Support Vector Machines for Diagnosis of Breast Tumors on US Images,” Academic Radiology, vol. 10(2), pp. 189-197, 2003.
[33] 鄭慶明，《實用影像診斷學》，初版，臺北市，俊傑，民國九十年。
[34] Gonzalez, R. C.與Woods, R. E.著，《數位影像處理》，繆紹綱譯，臺北市，臺灣培生教育，民國九十二年。
[35] 鍾國亮，《影像處理與電腦視覺》，四版，臺北市，臺灣東華，民國九十七年。
[36] 陳順宇，《多變量分析》，四版，臺北市，華泰，民國九十四年。
[37] 王雪松，程玉虎，《機器學習理論、方法與應用》，初版，北京，科學出版社，民國九十八年。
[38] Kindt, T. J.，Goldsby, R. A.與Osborne, B. A.著，《Kuby免疫學》，高榮駿譯，六版，臺北市，九州圖書，民國一百年。
[39] 莫宏偉，左興權，《人工免疫系統》，初版，北京，科學出版社，民國九十八年。
[40] 肖人彬，曹鵬彬，劉勇，《工程免疫計算》，初版，北京，科學出版社，民國九十六年。
[41] 焦李成，杜海峰，劉芳，公茂果，《免疫優化計算、學習與識別》，初版，北京，科學出版社，民國九十五年。
[42] 行政院衛生署：99年國人主要死因統計(以ICD-10編碼) http://www.doh.gov.tw/CHT2006/DM/DM2_p01.aspx?class_no=25&level_no=1&doc_no=80728 (accessed on July 11, 2011)
[43] 行政院衛生署：乳癌整體5年存活率可達8.5成 http://www.doh.gov.tw/CHT2006/DM/DM2_p01.aspx?class_no=25&level_no=1&doc_no=77127 (accessed on July 11, 2011)