本研究建立一個乳房X光影像辨識系統。在分類模型方面，使用卷積神經網路進行訓練與分類。而對模型過度擬合問題的抑制，使用資料增強、隨機關閉神經元以及批量標準化等方法。通常來說，模型性能與其結構息息相關，因此使用貝氏最佳化方法去找尋最佳化的模型結構。本研究使用mini-MIAS資料庫的乳房X光影像作為結構最佳化的驗證與分析，並以異常影像為正樣本，正常影像為負樣本。實驗結果顯示，由貝氏最佳化搜尋到的最佳結構模型，在測試資料的分類表現，真陽率為53.85%，假陽率為19.05%，準確率為70.59%。與參考模型的分類表現相比，有明顯的改善。

In this study, a system to detect breast abnormalities in mammograms is established. For the image classification, the convolutional neural network (CNN) is used for training and classification. Besides, to suppress model overfitting, data augmentation, dropout of neural network model, and batch normalization are utilized. Generally speaking, the model performance is closely related to its architecture; therefore, Bayesian optimization method is used to search the optimal model architecture. To validate and analyze the optimal model, mammograms from Mini-mammographic Image Analysis Society (MIAS) database are used in this study. In particular, abnormal cases are set as positive samples and normal cases are set as negative samples. Experimental results show that the best architectural model searched by Bayesian optimization yields 53.85% true positive rate, 19.05% false positive rate, and 70.59% accuracy rate for image classification in the test data. Compared with the referenced model, the improvement is significant.

[1] International Agency for Research on Cancer. World cancer day: Breast cancer overtakes lung cancer in terms of number of new cancer cases worldwide. https://www.iarc.who.int/news-events/world-cancer-day-2021/. Accessed 07,4,2022.
[2] 第三科癌症診療及照護。. 108年癌症登記報告。. https://www.hpa.gov.tw/Pages/Detail.aspx?nodeid=614&pid=1124. Accessed 07,4,2022.
[3] 翁文龠。. 乳癌早治療九成可痊癒。. http://www.cthyh.org.tw/?aid=304&pid=0&page_name=detail&iid=224. Accessed 07,4,2022.
[4] 衛生福利部國民健康署癌症防治組。. 乳癌防治。. https://www.hpa.gov.tw/Pages/Detail.aspx?nodeid=4576&pid=15750. Accessed 07,4,2022.
[5] 陳若瑀。. 認識乳房攝影。. https://epaper.ntuh.gov.tw/health/201501/special_1_2.html. Accessed 07,4,2022.
[6] Zhiqiong Wang, Ge Yu, Yan Kang, Yingjie Zhao, and Qixun Qu. Breast tumor detection in digital mammography based on extreme learning machine. Neurocomputing, 128:175–184, 2014.
[7] Hyun I Kim, Sung Shin, WeiWang, and Soon I Jeon. Svm-based harris corner detection for breast mammogram image normal/abnormal classification. In Proceedings of the 2013 Research in Adaptive and Convergent Systems, pages 187–191. 2013.
[8] Min Dong, Zhe Wang, Chenghui Dong, Xiaomin Mu, and Yide Ma. Classification of region of interest in mammograms using dual contourlet transform and improved knn. Journal of Sensors, 2017, 2017.
[9] Yu-Dong Zhang, Chichun Pan, Xianqing Chen, and Fubin Wang. Abnormal breast identification by nine-layer convolutional neural network with parametric rectified linear unit and rank-based stochastic pooling. Journal of computational science, 27:57–68, 2018.
[10] YJ Tan, KS Sim, and FF Ting. Breast cancer detection using convolutional neural networks for mammogram imaging system. In 2017 International Conference on Robotics, Automation and Sciences (ICORAS), pages 1–5. IEEE, 2017.
[11] James Bergstra and Yoshua Bengio. Random search for hyper-parameter optimization. Journal of machine learning research, 13(2), 2012.
[12] Carl Edward Rasmussen. Gaussian processes in machine learning. In Summer school on machine learning, pages 63–71. Springer, 2003.
[13] Yann LeCun, L´eon Bottou, Yoshua Bengio, and Patrick Haffner. Gradientbased learning applied to document recognition. Proceedings of the IEEE, 86(11):2278–2324, 1998.
[14] Kumar Chellapilla, Sidd Puri, and Patrice Simard. High performance convolutional neural networks for document processing. In Tenth international workshop on frontiers in handwriting recognition. Suvisoft, 2006.
[15] Yorick Hardy and Willi-hans Steeb. Problems and solutions in introductory and advanced matrix calculus. World Scientific Publishing Company, 2016.
[16] Sebastian Raschka and Vahid Mirjalili. Python 機器學習(Python Machine Learning)。劉立博碩文化, 2018.
[17] Ning Qian. On the momentum term in gradient descent learning algorithms. Neural networks, 12(1):145–151, 1999.
[18] Tijmen Tieleman, Geoffrey Hinton, et al. Lecture 6.5-rmsprop: Divide the gradient by a running average of its recent magnitude. COURSERA: Neural networks for machine learning, 4(2):26–31, 2012.
[19] Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014.
[20] Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, Ilya Sutskever, and Ruslan Salakhutdinov. Dropout: a simple way to prevent neural networks from overfitting. The journal of machine learning research, 15(1):1929–1958, 2014.
[21] Sergey Ioffe and Christian Szegedy. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In International conference on machine learning, pages 448–456. PMLR, 2015.
[22] Tzu-Kuo Huang et al. A technical introduction to gaussian process regression.2006.
[23] Dennis D Cox and Susan John. A statistical method for global optimization. In [Proceedings] 1992 IEEE International Conference on Systems, Man, and Cybernetics, pages 1241–1246. IEEE, 1992.
[24] Harold J Kushner. A new method of locating the maximum point of an arbitrary multipeak curve in the presence of noise. 1964.
[25] Eric Brochu, Vlad M Cora, and Nando De Freitas. A tutorial on bayesian optimization of expensive cost functions, with application to active user modeling and hierarchical reinforcement learning. arXiv preprint arXiv:1012.2599, 2010.
[26] Jonas Mockus, Vytautas Tiesis, and Antanas Zilinskas. The application of bayesian methods for seeking the extremum. Towards global optimization, 2(117-129):2, 1978.
[27] Francois Chollet et al. Keras, 2015.
[28] The GPyOpt authors. GPyOpt: A bayesian optimization framework in python. http://github.com/SheffieldML/GPyOpt, 2016.
[29] J.Suckling et al. The mammographic image analysis society digital mammogram database. International Congress Series, 1069:375–378, 1994.
[30] Shuyue Guan and Murray Loew. Breast cancer detection using transfer learning in convolutional neural networks. In 2017 IEEE Applied Imagery Pattern Recognition Workshop (AIPR), pages 1–8. IEEE, 2017.
[31] Inˆes C Moreira, Igor Amaral, Inˆes Domingues, Ant´onio Cardoso, Maria Joao Cardoso, and Jaime S Cardoso. Inbreast: toward a full-field digital mammographic database. Academic radiology, 19(2):236–248, 2012.
[32] Rebecca Sawyer Lee, Francisco Gimenez, Assaf Hoogi, Kanae Kawai Miyake, Mia Gorovoy, and Daniel L Rubin. A curated mammography data set for use in computer-aided detection and diagnosis research. Scientific data, 4(1):1–9, 2017.
[33] WilliamWHager. Updating the inverse of a matrix. SIAM review, 31(2):221– 239,1989.