本論文使用弱監督學習方式，針對全玻片病理影像(WSI)開發一電腦輔助診斷分析系統對腦腫瘤進行分型分類。訓練集中包含了221例全玻片腦腫瘤病理影像，並分類成三個分型: 星狀細胞瘤(Astrocytoma)、多形性膠質母細胞瘤(Glioblastoma Multiform) 與寡樹突膠質瘤(Oligodendroglioma)，但卻不包含各類別在病理圖像中所對應病灶區域的標註。 另外，測試集有35例用於演算法上的效能評估。有別於以往的方法需要先由醫學專家對針對病理影像上的病灶區域進行標註，再對所標註的病灶區域進行採樣以收集訓練集來訓練深度模型；本研究中的實驗只提供每例病理影像的腦腫瘤分型標註，而未有該腫瘤分型對應的病灶區域標註。比較可行的方法便是對所有的區域隨機採樣，並將所有從該例WSI中的採樣樣本都標註與該例腫瘤分型標註相同。然而，並非所有的採樣皆為該腫瘤分型的代表病灶區域。如此一來，弱監督下的不確切標註會產生錯誤的雜訊標註，導致訓練模型受到錯誤標註的影響而無法學習到各類病灶的真實特性。因此，本論文著重於耐高雜訊的深度學習方法以突破弱監督下的限制。由於錯誤標註的比例無法預先得，prototypes代表選取是較為適用的方法以降低錯誤標註對分類結果的影響。藉由測試影像與各類prototypes的特徵比對方法進行分類的準確率可從原本直接使用模型的0.68 升高至 0.77。此外，使用prototypes代表選取的另一個目的是為了大幅降低專家介入的人力成本以減輕病理醫師的負擔。也就是說，醫事專業人員只需針對各病灶分型所選出的prototypes 進行檢查與修正錯誤標註即可，而不用對所有的訓練集中的採樣影像一一檢查。藉由領域專家的少許人為介入更正錯誤分類與去除與分類無關的prototypes後，分類效能的良率又進一步從0.77 提升至 0.86。實驗結果證實藉由prototypes代表選取方法能有效降低錯誤標註的干擾，且藉由少許的專家介入便能進一步移除大部分的錯誤標註，達到大幅降低人力與時間成本的功效。

This thesis presents a computer-aided diagnosis (CAD) system for brain tumor classification based on pathological imaging under very limited supervision. The total number of 221 whole-slide images (WSIs) were classified into three categories: Astrocytoma, Glioblastoma Multiform (GBM), and Oligodendroglioma for training without the precise lesion annotations given, and the testing set that contains 35 cases were used for performance evaluation. Different from the conventional approaches in deep learning that the deep models were trained on the dataset that can be obtained by sampling patches based on the annotations from medical experts, only the cases’ labels were given in this task without knowing the lesions’ positions. All sampling patches were labeled based on their corresponding case categories. As a result, the inexact labelling in the task of weak supervision leads to the inaccurate samples in the training process. Therefore, the paper focuses on the deep learning approaches that are featured with higher noise-tolerance to overcome the limitations. Since the ratio of noisy samples was unknown and unpredictable, the method of prototype selection is suited for the task to reduce the impact from those inaccurate labels. The classification performance was raised from 0.68 to 0.77 by feature mapping with the selected prototypes of each class. Furthermore, another purpose of such prototype approach is to largely reduce the level of expert intervention to alleviate the load from pathologists. That is, the medical experts only need to examine the prototypes sets for label correction instead of checking all the sampling patches. The performance of classification can be further improved from 0.77 to 0.86 with such fashion and the manual correction can be achieved much easier.

[1] 衛生福利部國民健康署, "中華民國 106 年癌症登記報告," 2019.
[2] 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.
[3] M. D. Zeiler and R. Fergus, "Visualizing and understanding convolutional networks," in European conference on computer vision, 2014: Springer, pp. 818-833.
[4] J. Long, E. Shelhamer, and T. Darrell, "Fully convolutional networks for semantic segmentation," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 3431-3440.
[5] M.-C. Popescu, V. E. Balas, L. Perescu-Popescu, and N. Mastorakis, "Multilayer perceptron and neural networks," WSEAS Transactions on Circuits and Systems, vol. 8, no. 7, pp. 579-588, 2009.
[6] R. Pascanu, T. Mikolov, and Y. Bengio, "On the difficulty of training recurrent neural networks," in International conference on machine learning, 2013: PMLR, pp. 1310-1318.
[7] V. Nair and G. E. Hinton, "Rectified linear units improve restricted boltzmann machines," in Icml, 2010.
[8] D.-A. Clevert, T. Unterthiner, and S. Hochreiter, "Fast and accurate deep network learning by exponential linear units (elus)," arXiv preprint arXiv:1511.07289, 2015.
[9] W. Shang, K. Sohn, D. Almeida, and H. Lee, "Understanding and improving convolutional neural networks via concatenated rectified linear units," in international conference on machine learning, 2016: PMLR, pp. 2217-2225.
[10] G. Klambauer, T. Unterthiner, A. Mayr, and S. Hochreiter, "Self-normalizing neural networks," arXiv preprint arXiv:1706.02515, 2017.
[11] B. Xu, N. Wang, T. Chen, and M. Li, "Empirical evaluation of rectified activations in convolutional network," arXiv preprint arXiv:1505.00853, 2015.
[12] X. Glorot, A. Bordes, and Y. Bengio, "Deep sparse rectifier neural networks," in Proceedings of the fourteenth international conference on artificial intelligence and statistics, 2011: JMLR Workshop and Conference Proceedings, pp. 315-323.
[13] C. Gulcehre, M. Moczulski, M. Denil, and Y. Bengio, "Noisy activation functions," in International conference on machine learning, 2016: PMLR, pp. 3059-3068.
[14] M. D. Zeiler and R. Fergus, "Stochastic pooling for regularization of deep convolutional neural networks," arXiv preprint arXiv:1301.3557, 2013.
[15] C. Gulcehre, K. Cho, R. Pascanu, and Y. Bengio, "Learned-norm pooling for deep feedforward and recurrent neural networks," in Joint European Conference on Machine Learning and Knowledge Discovery in Databases, 2014: Springer, pp. 530-546.
[16] S. Ruder, "An overview of gradient descent optimization algorithms," arXiv preprint arXiv:1609.04747, 2016.
[17] I. Sutskever, J. Martens, G. Dahl, and G. Hinton, "On the importance of initialization and momentum in deep learning," in International conference on machine learning, 2013: PMLR, pp. 1139-1147.
[18] A. Botev, G. Lever, and D. Barber, "Nesterov's accelerated gradient and momentum as approximations to regularised update descent," in 2017 International Joint Conference on Neural Networks (IJCNN), 2017: IEEE, pp. 1899-1903.
[19] Ö. Çiçek, A. Abdulkadir, S. S. Lienkamp, T. Brox, and O. Ronneberger, "3D U-Net: learning dense volumetric segmentation from sparse annotation," in International conference on medical image computing and computer-assisted intervention, 2016: Springer, pp. 424-432.
[20] D. P. Kingma and J. Ba, "Adam: A method for stochastic optimization," arXiv preprint arXiv:1412.6980, 2014.
[21] C. Shorten and T. M. Khoshgoftaar, "A survey on image data augmentation for deep learning," Journal of Big Data, vol. 6, no. 1, pp. 1-48, 2019.
[22] A. Krizhevsky, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks," Advances in neural information processing systems, vol. 25, pp. 1097-1105, 2012.
[23] J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, "Imagenet: A large-scale hierarchical image database," in 2009 IEEE conference on computer vision and pattern recognition, 2009: Ieee, pp. 248-255.
[24] G. H. Dunteman, Principal components analysis (no. 69). Sage, 1989.
[25] N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, and R. Salakhutdinov, "Dropout: a simple way to prevent neural networks from overfitting," The journal of machine learning research, vol. 15, no. 1, pp. 1929-1958, 2014.
[26] K. Simonyan and A. Zisserman, "Very deep convolutional networks for large-scale image recognition," arXiv preprint arXiv:1409.1556, 2014.
[27] C. Szegedy et al., "Going deeper with convolutions," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 1-9.
[28] 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.
[29] M. Abadi et al., "Tensorflow: A system for large-scale machine learning," in 12th {USENIX} symposium on operating systems design and implementation ({OSDI} 16), 2016, pp. 265-283.
[30] Y. Jia et al., "Caffe: Convolutional architecture for fast feature embedding," in Proceedings of the 22nd ACM international conference on Multimedia, 2014, pp. 675-678.
[31] A. Paszke et al., "Pytorch: An imperative style, high-performance deep learning library," arXiv preprint arXiv:1912.01703, 2019.
[32] T.-Y. Lin et al., "Microsoft coco: Common objects in context," in European conference on computer vision, 2014: Springer, pp. 740-755.
[33] C. G. Northcutt, L. Jiang, and I. L. Chuang, "Confident learning: Estimating uncertainty in dataset labels," arXiv preprint arXiv:1911.00068, 2019.
[34] B. Han et al., "Co-teaching: Robust training of deep neural networks with extremely noisy labels," arXiv preprint arXiv:1804.06872, 2018.
[35] K.-H. Lee, X. He, L. Zhang, and L. Yang, "Cleannet: Transfer learning for scalable image classifier training with label noise," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 5447-5456.
[36] G. Guo, H. Wang, D. Bell, Y. Bi, and K. Greer, "KNN model-based approach in classification," in OTM Confederated International Conferences" On the Move to Meaningful Internet Systems", 2003: Springer, pp. 986-996.
[37] J. Han, P. Luo, and X. Wang, "Deep self-learning from noisy labels," in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 5138-5147.
[38] T. Xiao, T. Xia, Y. Yang, C. Huang, and X. Wang, "Learning from massive noisy labeled data for image classification," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 2691-2699.
[39] X. Ma and F. Jia, "Brain Tumor Classification with Multimodal MR and Pathology Images," in International MICCAI Brainlesion Workshop, 2019: Springer, pp. 343-352.
[40] T. D. Bui, J. Shin, and T. Moon, "3D densely convolutional networks for volumetric segmentation," arXiv preprint arXiv:1709.03199, 2017.
[41] L. Pei, L. Vidyaratne, W.-W. Hsu, M. M. Rahman, and K. M. Iftekharuddin, "Brain Tumor Classification Using 3D Convolutional Neural Network," in International MICCAI Brainlesion Workshop, 2019: Springer, pp. 335-342.
[42] H.-W. Chan, Y.-T. Weng, and T.-Y. Huang, "Automatic Classification of Brain Tumor Types with the MRI Scans and Histopathology Images," in International MICCAI Brainlesion Workshop, 2019: Springer, pp. 353-359.
[43] T. Kanungo, D. M. Mount, N. S. Netanyahu, C. D. Piatko, R. Silverman, and A. Y. Wu, "An efficient k-means clustering algorithm: Analysis and implementation," IEEE transactions on pattern analysis and machine intelligence, vol. 24, no. 7, pp. 881-892, 2002.
[44] L. Breiman, "Random forests," Machine learning, vol. 45, no. 1, pp. 5-32, 2001.