依據世界衛生組織公布之資料，肺癌為現今致死率最高的癌症，每年約有一百六十萬人因其致死。肺癌可細分為諸多類別且各有不同的治療方式，例如：在肺腺癌的治療中，若先進行基因檢測，確認基因突變的種類，則有機會使用適當的標靶藥物，抑制腫瘤生長延長病患壽命。先前對於醫療影像的分析多採用基於專業背景知識或進階影像技術所提取的特徵來進行分析，但要找到足以正確描述大量影像的特徵實屬不易。近年來，發展快速的神經網路有自行學習影像特徵的能力，且在影像辨識方面已有顯著的結果，搭配硬體上圖形處理器的加速，使得深度學習得以多方面應用。
本研究主要分為兩大部分，分別為影像辨識與深度學習模型分析。在辨識部分，我們採用深度學習技術，使用205張全切片肺癌影像，針對「是否為腫瘤組織」與「是否為表皮生長因子接受器（epidermal growth factor receptor, EGFR）基因突變」二種目標分別訓練分類模型，建置一個高準確度肺癌病理影像辨識系統，快速識別出癌症組織與肺腺癌特定基因突變種類，其正確率分別可達到96.43%和97.63%。在分析的部分，深度模型雖然具有極佳的準確率，但對於模型判斷的方式卻缺乏明確解釋，我們欲找出深度模型用以進行分類的特徵，並採用風格轉換及統計影像特徵值，來進行模型的驗證與解釋。

According to the World Health Organization, lung cancer is the leading cause of cancer death, resulting in about 1.6 million deaths each year. There exist many subtypes of lung cancer, and each of them has a different treatment option. For example, if the type of genetic mutation can be confirmed in the treatment of lung adenocarcinoma, targeted therapy can be adopted to block the driver mutation for inhibiting cancer cell growth.
In this study, we trained two convolution neural networks on 205 whole-slide images for lung cancer classification. One is to classify an input patch as tumor or normal, and the other is to predict whether the patch is the epidermal growth factor receptor (EGFR) gene mutation or not. These models achieved the highest classification rates of 96.43% and 97.63%, respectively. Furthermore, we try to explore the features extracted by these deep learning models, where the style transfer and feature statistics are used for their interpretation. The experimental results demonstrate that our developed methods can assist in the classification of lung cancer whole-slide images and gives insights for deep learning models.

[1] F. L. Greene et al., AJCC cancer staging manual. Springer, 2002.
[2] L. Horn, C. Lovly, and D. Johnson, "Chapter 107: Neoplasms of the lung," Kasper, DL; Hauser, SL; Jameson, JL; Fauci, AS, 2015.
[3] J. Ferlay et al., "Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012," International journal of cancer, vol. 136, no. 5, pp. E359-E386, 2015.
[4] F. Bray, J. Ferlay, I. Soerjomataram, R. L. Siegel, L. A. Torre, and A. Jemal, "Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries," CA: a cancer journal for clinicians, vol. 68, no. 6, pp. 394-424, 2018.
[5] B. Stewart and C. P. Wild, "World cancer report 2014," 2014.
[6] L. G. Collins, C. Haines, R. Perkel, and R. E. Enck, "Lung cancer: diagnosis and management," Am Fam Physician, vol. 75, no. 1, pp. 56-63, 2007.
[7] W. D. Travis et al., "International association for the study of lung cancer/american thoracic society/european respiratory society international multidisciplinary classification of lung adenocarcinoma," Journal of thoracic oncology, vol. 6, no. 2, pp. 244-285, 2011.
[8] B. A. Chan and B. G. Hughes, "Targeted therapy for non-small cell lung cancer: current standards and the promise of the future," Translational lung cancer research, vol. 4, no. 1, p. 36, 2015.
[9] U. Djuric, G. Zadeh, K. Aldape, and P. Diamandis, "Precision histology: how deep learning is poised to revitalize histomorphology for personalized cancer care," NPJ precision oncology, vol. 1, no. 1, p. 22, 2017.
[10] N. Stathonikos, M. Veta, A. Huisman, and P. J. van Diest, "Going fully digital: Perspective of a Dutch academic pathology lab," Journal of pathology informatics, vol. 4, 2013.
[11] Y. Shi et al., "A prospective, molecular epidemiology study of EGFR mutations in Asian patients with advanced non–small-cell lung cancer of adenocarcinoma histology (PIONEER)," Journal of thoracic oncology, vol. 9, no. 2, pp. 154-162, 2014.
[12] K. Tomczak, P. Czerwińska, and M. Wiznerowicz, "The Cancer Genome Atlas (TCGA): an immeasurable source of knowledge," Contemporary oncology, vol. 19, no. 1A, p. A68, 2015.
[13] D. Shen, G. Wu, and H.-I. Suk, "Deep learning in medical image analysis," Annual review of biomedical engineering, vol. 19, pp. 221-248, 2017.
[14] F. Xing, Y. Xie, and L. Yang, "An automatic learning-based framework for robust nucleus segmentation," IEEE transactions on medical imaging, vol. 35, no. 2, pp. 550-566, 2015.
[15] T. Qaiser, Y.-W. Tsang, D. Epstein, and N. Rajpoot, "Tumor segmentation in whole slide images using persistent homology and deep convolutional features," in Annual Conference on Medical Image Understanding and Analysis, 2017, pp. 320-329: Springer.
[16] M. Veta et al., "Prognostic value of automatically extracted nuclear morphometric features in whole slide images of male breast cancer," Modern pathology, vol. 25, no. 12, p. 1559, 2012.
[17] F. Dong et al., "Computational pathology to discriminate benign from malignant intraductal proliferations of the breast," PloS one, vol. 9, no. 12, p. e114885, 2014.
[18] F. Dong et al., "Computational pathology to discriminate benign from malignant intraductal proliferations of the breast," vol. 9, no. 12, p. e114885, 2014.
[19] J. N. Kather et al., "Multi-class texture analysis in colorectal cancer histology," Scientific reports, vol. 6, p. 27988, 2016.
[20] W. S. McCulloch and W. Pitts, "A logical calculus of the ideas immanent in nervous activity," Bulletin of mathematical biology, vol. 52, no. 1-2, pp. 99-115, 1990.
[21] Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, "Gradient-based learning applied to document recognition," Proceedings of the IEEE, vol. 86, no. 11, pp. 2278-2324, 1998.
[22] P. J. P. D. d. Werbos, Harvard University, "Beyond Regression:" New Tools for Prediction and Analysis in the Behavioral Sciences," 1974.
[23] 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.
[24] A. Krizhevsky, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks," in Advances in neural information processing systems, 2012, pp. 1097-1105.
[25] K. Simonyan and A. Zisserman, "Very deep convolutional networks for large-scale image recognition," arXiv preprint arXiv:1409.1556, 2014.
[26] C. Szegedy et al., "Going deeper with convolutions," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 1-9.
[27] K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770-778.
[28] L. Hou, D. Samaras, T. M. Kurc, Y. Gao, J. E. Davis, and J. H. Saltz, "Patch-based convolutional neural network for whole slide tissue image classification," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 2424-2433.
[29] N. Coudray et al., "Classification and mutation prediction from non–small cell lung cancer histopathology images using deep learning," Nature medicine, vol. 24, no. 10, p. 1559, 2018.
[30] S. Graham, M. Shaban, T. Qaiser, S. A. Khurram, and N. Rajpoot, "Classification of lung cancer histology images using patch-level summary statistics," in Medical Imaging 2018: Digital Pathology, 2018, vol. 10581, p. 1058119: International Society for Optics and Photonics.
[31] P. Bankhead et al., "QuPath: Open source software for digital pathology image analysis," Scientific reports, vol. 7, no. 1, p. 16878, 2017.
[32] D. Komura and S. Ishikawa, "Machine learning methods for histopathological image analysis," Computational and structural biotechnology journal, vol. 16, pp. 34-42, 2018.
[33] D. Wang, A. Khosla, R. Gargeya, H. Irshad, and A. H. Beck, "Deep learning for identifying metastatic breast cancer," arXiv preprint arXiv:1606.05718, 2016.
[34] B. Xu, N. Wang, T. Chen, and M. Li, "Empirical evaluation of rectified activations in convolutional network," arXiv preprint arXiv:1505.00853, 2015.
[35] C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, "Rethinking the inception architecture for computer vision," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 2818-2826.
[36] S. J. Pan and Q. Yang, "A survey on transfer learning," IEEE Transactions on knowledge and data engineering, vol. 22, no. 10, pp. 1345-1359, 2010.
[37] L. A. Gatys, A. S. Ecker, and M. Bethge, "A neural algorithm of artistic style," arXiv preprint arXiv:1508.06576, 2015.
[38] L. A. Gatys, A. S. Ecker, and M. Bethge, "Texture synthesis and the controlled generation of natural stimuli using convolutional neural networks," in Bernstein Conference 2015, 2015, pp. 219-219.
[39] A. E. Carpenter et al., "CellProfiler: image analysis software for identifying and quantifying cell phenotypes," Genome biology, vol. 7, no. 10, p. R100, 2006.
[40] X. Luo et al., "Comprehensive computational pathological image analysis predicts lung cancer prognosis," Journal of Thoracic Oncology, vol. 12, no. 3, pp. 501-509, 2017.
[41] 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.
[42] Z. von F, "Beugungstheorie des schneidenver-fahrens und seiner verbesserten form, der phasenkontrastmethode," Physica, vol. 1, no. 7-12, pp. 689-704, 1934.
[43] Y.-W. Chen and C.-J. Lin, "Combining SVMs with various feature selection strategies," in Feature extraction: Springer, 2006, pp. 315-324.