乳癌是台灣女性因癌症死亡的主要原因之一，也是全球發病人口最多的癌症種類，而在乳癌早期篩檢出病灶並接受治療能有效地降低其死亡率。乳房Ｘ光攝影是台灣目前用來進行篩檢乳癌早期病灶之微鈣化群的常見方法之一，但是因為微鈣化的病灶特徵不易觀察，所以如何幫助放射科醫生能盡快且有效地找出微鈣化群區域將是一個重要的議題。
在本文中，提出一種使用微鈣化群標記資料的自動預測病灶與圖像分割的方法，使用U-Net及V-Net兩種卷積網路架構進行比較，並且在卷積網路架構中結合標準化、Dropout及以BCE結合Dice做為損失函數用於對微鈣化群的偵測，且使用了其他的預處理方法以求達到最佳的檢測結果，其中包括窗口調整、影像預處理、資料增生。最終，將乳房Ｘ光圖像以訓練好的模型對微鈣化群進行病灶預測及圖像分割，以輔助放射科醫生進行診斷。
私有數據集上的實驗結果顯示，U-Net在兩種卷積網路架構中取得最佳結果，在使用原始圖片進行訓練的結果，其準確率達到72%，使用影像預處理後的圖片進行訓練的結果達到81%的準確率，而使用複合方法搭配影像預處理後的圖片進行訓練的結果達到85%的準確率。V-Net使用影像預處理後的圖片進行訓練之結果僅達到70%的準確率，而使用複合方法搭配影像預處理後的圖片進行訓練之結果僅達到72%的準確率。

Breast cancer is the most common type of cancer in the world and the leading cause of cancer-related death in Taiwanese women. However, the way to reduce breast cancer mortality is to be able to detect lesions and treat them at an early stage. Mammography is a universally effective method for screening for calcification in early breast cancer lesions in Taiwan, but the characteristics of micro-calcifications are difficult to observe. Therefore, how to help radiologists find micro-calcification clusters as soon as possible and effectively is an important topic.

In this paper, we propose an automatic lesion prediction and image segmentation method using micro-calcification cluster labeled data, We compared two convolutional network architectures, U-Net and V-Net. Using normalization, dropout, and BCE combined with Dice as a loss function to detect micro-calcification clusters in convolutional network architectures, among others Preprocessing method for best results. Other preprocessing methods include window resizing, image preprocessing, and data augmentation. Finally, the trained model was used to predict lesions and image segmentation of the micro-calcification group to aid radiologists in their diagnosis.

The experimental results on the private datasets show U-Net achieves the best results among the two convolutional network architectures. When trained with the original images have an accuracy rate of 72%, the result of training with images after image preprocessing reaches 81% accuracy, the results of training using the composite method with preprocessed images reach 85% accuracy. The training result of V-Net using preprocessed images only reached 70% accuracy, while the result of training using the composite method with preprocessed images only achieved 72% accuracy.

[1]Sung, H., et al., "Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries". CA: a cancer journal for clinicians. 71(3): pp. 209-249, 2021.
[2]衛生福利部。109年死因統計結果分析。衛生福利部統計處，pp. 2–7,2020.
[3]衛生福利部。乳癌防治。衛生福利部國民健康署癌症防治組，2022.https://www.hpa.gov.tw/Pp./Detail.aspx?nodeid=614&pid=1124
[4] Pardamean, B., et al., "Transfer learning from chest X-ray pre-trained convolutional neural network for learning mammogram data". Procedia Computer Science. 135: pp. 400-407, 2018.
[5]黃建銓。乳癌攝影篩檢制度之統計議題。國立高雄大學統計學研究所碩士論文，2010。
[6]Lehman, C.D., et al., "Diagnostic accuracy of digital screening mammography with and without computer-aided detection". JAMA internal medicine. 175(11): pp. 1828-1837, 2015.
[7]Khalaf, A.F. and I.A. Yassine, "Novel features for microcalcification detection in digital mammogram images based on wavelet and statistical analysis". in 2015 IEEE International Conference on Image Processing (ICIP). pp. 1825-1829, 2015.
[8]John McCarthy,"WHAT IS ARTIFICIAL INTELLIGENCE?".Education,1998.
[9]Turing, A.M., "Computing machinery and intelligence (1950)". The Essential Turing: the Ideas That Gave Birth to the Computer Age: pp. 433-464, 2012.
[10]Yang, X., et al., "Concepts of artificial intelligence for computer-assisted drug discovery". Chemical reviews. 119(18): pp. 10520-10594, 2019.
[11]Alpaydin, E,"Introduction to Machine Learning".1th Edition, The MIT Press, Cambridge,2004.
[12]Schapire, R.E., "The strength of weak learnability". Machine learning. 5(2): pp. 197-227, 1990.
[13]Freund, Y., "Boosting a weak learning algorithm by majority". Information and computation. 121(2): pp. 256-285, 1995.
[14]Cortes, C. and V. Vapnik, "Support-vector networks". Machine learning. 20(3): pp. 273-297, 1995.
[15]Freund, Y. and R.E. Schapire, "A decision-theoretic generalization of on-line learning and an application to boosting". Journal of computer and system sciences. 55(1): pp. 119-139, 1997.
[16]Breiman, L., "Random forests". Machine learning. 45(1): pp. 5-32, 2001.
[17]Homayounpour, M.M. and P.A. SOLTANI, "Speech acts classification of Persian language texts using three machine learning methods". pp. 65-71, 2010.
[18]Sellam, V. and J. Jagadeesan, "Classification of normal and pathological voice using SVM and RBFNN". Journal of signal and Information Processing. 5(1): pp. 1–7, 2014.
[19]Nugrahaeni, R.A. and K. Mutijarsa, "Comparative analysis of machine learning KNN, SVM, and random forests algorithm for facial expression classification". in 2016 International Seminar on Application for Technology of Information and Communication (ISemantic). pp. 163-168, 2016.
[20]Machhale, K., et al, "MRI brain cancer classification using hybrid classifier (SVM-KNN)". in 2015 International Conference on Industrial Instrumentation and Control (ICIC). pp. 60-65, 2015.
[21]Kumar, M. and M. Thenmozhi, "Forecasting stock index movement: A comparison of support vector machines and random forest". in Indian institute of capital markets 9th capital markets conference paper. 2006.
[22]McCulloch, W.S. and W. Pitts, "A logical calculus of the ideas immanent in nervous activity". The bulletin of mathematical biophysics. 5(4): pp. 115-133, 1943.
[23]Hebb, D.O,"The Organization of Behavior".New York: Wiley & Sons,1949.
[24]Rosenblatt, F., "The perceptron: a probabilistic model for information storage and organization in the brain". Psychological review. 65(6): pp. 386, 1958.
[25]Rumelhart, D.E., G.E. Hinton, and R.J. Williams, "Learning representations by back-propagating errors". nature. 323(6088): pp. 533-536, 1986.
[26]Salakhutdinov, R., A. Mnih, and G. Hinton, "Restricted Boltzmann machines for collaborative filtering". in Proceedings of the 24th international conference on Machine learning. pp. 791-798, 2007.
[27]Xu, L., et al., "Deep convolutional neural network for image deconvolution". Advances in neural information processing systems. 27: pp. 1790-1798, 2014.
[28]Cho, K., et al., "Learning phrase representations using RNN encoder-decoder for statistical machine translation". arXiv preprint arXiv:1406.1078,2014.
[29]Graves, A., A.-r. Mohamed, and G. Hinton, "Speech recognition with deep recurrent neural networks". in 2013 IEEE international conference on acoustics, speech and signal processing. pp. 6645-6649, 2013.
[30]Zamir, R., et al., "Segmenting microcalcifications in mammograms and its applications". in Medical Imaging 2021: Image Processing. pp. 788-795, 2021.
[31]Moghar, A. and M. Hamiche, "Stock market prediction using LSTM recurrent neural network". Procedia Computer Science. 170: pp. 1168-1173, 2020.
[32]Sainath, T.N., et al., "Improvements to deep convolutional neural networks for LVCSR". in 2013 IEEE workshop on automatic speech recognition and understanding. pp. 315-320, 2013.
[33]Bochkovskiy, A., C.-Y. Wang, and H.-Y.M. Liao, "Yolov4: Optimal speed and accuracy of object detection". arXiv preprint arXiv:2004.10934,2020.
[34]Karpathy, A., et al., "Large-scale video classification with convolutional neural networks". in Proceedings of the IEEE conference on Computer Vision and Pattern Recognition. pp. 1725-1732, 2014.
[35]Ding, J., et al., "Accurate pulmonary nodule detection in computed tomography images using deep convolutional neural networks". in International Conference on Medical Image Computing and Computer-Assisted Intervention. pp. 559-567, 2017.
[36]Yasaka, K., et al., "Deep learning with convolutional neural network for differentiation of liver masses at dynamic contrast-enhanced CT: a preliminary study". Radiology. 286(3): pp. 887-896, 2018.
[37]陳擁銘。基於多視野卷積神經網路之息肉狀脈絡膜血管病變偵測。國立中正大學資訊工程研究所碩士論文，2017。
[38]LeCun, Y., et al., "Gradient-based learning applied to document recognition". Proceedings of the IEEE. 86(11): pp. 2278-2324, 1998.
[39]Lin, T.-Y., et al., "Microsoft coco: Common objects in context". in European conference on computer vision. pp. 740-755, 2014.
[40]Kirillov, A., et al., "Panoptic segmentation". in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 9404-9413, 2019.
[41]Ronneberger, O., P. Fischer, and T. Brox, "U-net: Convolutional networks for biomedical image segmentation". in International Conference on Medical image computing and computer-assisted intervention. pp. 234-241, 2015.
[42]Kazemifar, S., et al., "Segmentation of the prostate and organs at risk in male pelvic CT images using deep learning". Biomedical Physics & Engineering Express. 4(5): pp. 055003, 2018.
[43]Nielsen, M.A., Neural networks and deep learning. Determination press San Francisco, CA, USA. Vol. 25. 2015.
[44]Haykin, S. and N. Network, "A comprehensive foundation". Neural networks. 2(2004): pp. 41, 2004.
[45]Glorot, X., A. Bordes, and Y. Bengio, "Deep sparse rectifier neural networks". in Proceedings of the fourteenth international conference on artificial intelligence and statistics. pp. 315-323, 2011.
[46]Cid-Sueiro, J., et al., "Cost functions to estimate a posteriori probabilities in multiclass problems". IEEE Transactions on Neural Networks. 10(3): pp. 645-656, 1999.
[47]Ketkar, N. and E. Santana, Deep learning with Python. Vol. 1. 2017.
[48]Milletari, F., N. Navab, and S.-A. Ahmadi, "V-net: Fully convolutional neural networks for volumetric medical image segmentation". in 2016 fourth international conference on 3D vision (3DV). pp. 565-571, 2016.
[49]Sun, H., et al., "AUNet: attention-guided dense-upsampling networks for breast mass segmentation in whole mammograms". Physics in Medicine & Biology. 65(5): pp. 055005, 2020.
[50]Srivastava, N., et al., "Dropout: a simple way to prevent neural networks from overfitting". The journal of machine learning research. 15(1): pp. 1929-1958, 2014.
[51]Ioffe, S. and C. Szegedy, "Batch normalization: Accelerating deep network training by reducing internal covariate shift". in International conference on machine learning. pp. 448-456, 2015.
[52]Shorten, C. and T.M. Khoshgoftaar, "A survey on image data augmentation for deep learning". Journal of big data. 6(1): pp. 1-48, 2019.
[53]Kamnitsas, K., et al., "Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation". Medical image analysis. 36: pp. 61-78, 2017.
[54]Cao, Z., et al., "Deeplima: Deep learning based lesion identification in mammograms". in Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops. pp. 362-370, 2019.
[55]Quintanilla-Domínguez, J., et al., "Image segmentation by fuzzy and possibilistic clustering algorithms for the identification of microcalcifications". Scientia Iranica. 18(3): pp. 580-589, 2011.
[56]黃國禎。全域數位乳房攝影之微鈣化群自動偵測系統。中原大學生物醫學工程學系研究所碩士論文，2008。
[57]Bazzani, A., et al., "An SVM classifier to separate false signals from microcalcifications in digital mammograms". Physics in Medicine & Biology. 46(6): pp. 1651, 2001.