非酒精性脂肪肝病（NAFLD）大致可分為兩種類型，非酒精性脂肪肝（NAFL）和非酒精性脂肪性肝炎（NASH）。在這兩種類型中，非酒精性脂肪肝炎更危險，其特徵是肝細胞膨脹。傳統上，NASH是由專業的病理學家在全景玻片影像（WSI）中手動檢測。然而，這個過程相當耗時且主觀容易出錯。因此，迫切需要電腦輔助診斷方法。因此，本文提出了一種模型自動檢測肝臟病理中氣球樣變性肝細胞，該模型基於U-net，添加了。(1) attention gates: 用以抑制模型學習不相關區域，並在訓練過程中同時加強相關特徵的學習；(2)recurrent units用以加深網絡結構，從而能夠在相同數量的模型參數下提取更好的特徵；(3) residual block用以加速訓練過程收斂。此外，本研究還提出了一種優化的主動學習算法，以解決訓練過程中數據不均衡的問題，並相應提高檢測性能。以台灣某一公立醫院62例NASH患者的註釋WSI為研究對象，研究了所提出模型和學習算法的可行性。結果表明，該方法可以檢測到氣球化變性肝細胞，F1-score為70.81%。此外，該方法還降低了約43%的標記成本。因此，該方法為醫生進行NASH病理檢查提供了一個有價值的輔助工具。

Non-alcoholic fatty liver disease (NAFLD) can be broadly classified into two types, namely non-alcoholic fatty liver (NAFL) and non-alcoholic steatohepatitis (NASH). Of the two types, NASH is more dangerous, and is characterised by hepatocellular ballooning. Traditionally, NASH is detected manually in whole slide images (WSIs) by a professional pathologist. However, the process is time-consuming, subjective, and prone to errors. Therefore, there is an urgent need for computer-aided diagnostic methods. Accordingly, this paper proposes a novel method for the automatic detection of ballooning cells in liver pathology based on a U-net model extended to include: (1) attention gates to suppress irrelevant areas of model learning and aggregate the learning of relevant features simultaneously during the training process; (2) recurrent units to deepen the network structure and therefore enable the extraction of better features with the same number of model parameters. and (3) residual units to accelerate the training process. The study additionally proposes an optimized active learning algorithm to solve the problem of data imbalance in the training process and improve the detection performance accordingly. The feasibility of the proposed model and learning algorithm is investigated using annotated WSIs obtained from 62 NASH patients at a public hospital in Taiwan. The results show that the method can detect ballooning cells with a F1-score of 70.81%. In addition, the method reduces the labeling cost by around 43%. The method thus provides a valuable assistive tool to doctors in performing NASH pathology.

[1] Z. M. Younossi, A. B. Koenig, D. Abdelatif, Y. Fazel, L. Henry, and M. Wymer, “Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes,” Hepatology, vol. 64, no. 1, pp. 73–84, 2016.
[2] Z. Younossi, F. Tacke, M. Arrese, B. Chander Sharma, I. Mostafa, E. Bugianesi, V. Wai-Sun Wong, Y. Yilmaz, J. George, J. Fan, and M. B. Vos, “Global Perspectives on Nonalcoholic Fatty Liver Disease and Nonalcoholic Steatohepatitis,” Hepatology, vol. 69, no. 6, pp. 2672–2682, 2019.
[3] C. Estes, Q. M. Anstee, M. T. Arias-Loste, H. Bantel, S. Bellentani, J. Caballeria, M. Colombo, A. Craxi, J. Crespo, C. P. Day, Y. Eguchi, A. Geier, L. A. Kondili, D. C. Kroy, J. V. Lazarus, R. Loomba, M. P. Manns, G. Marchesini, A. Nakajima, F. Negro, S. Petta, V. Ratziu, M. Romero-Gomez, A. Sanyal, J. M. Schattenberg, F. Tacke, J. Tanaka, C. Trautwein, L. Wei, S. Zeuzem, and H. Razavi, “Modeling NAFLD disease burden in China, France, Germany, Italy, Japan, Spain, United Kingdom, and United States for the period 2016–2030,” Journal of Hepatology, vol. 69, no. 4, pp. 896–904, 2018.
[4] Ludwig J, Viggiano TR, McGill DB, et al, “Nonalcoholic steatohepatitis: Mayo Clinic experiences with a hitherto unnamed disease,” Mayo Clin Proc 1980;55:434–438.
[5] E. Buzzetti, M. Pinzani, and E. A. Tsochatzis, “The multiple-hit pathogenesis of non-alcoholic fatty liver disease (NAFLD),” Metabolism, vol. 65, no. 8, pp. 1038–1048, 2016.
[6] J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic segmentation,” 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015.
[7] O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional Networks for Biomedical Image Segmentation,” Lecture Notes in Computer Science, pp. 234–241, 2015.
[8] M. Khened, V. A. Kollerathu, and G. Krishnamurthi, “Fully convolutional multi-scale residual DenseNets for cardiac segmentation and automated cardiac diagnosis using ensemble of classifiers,” Medical Image Analysis, vol. 51, pp. 21–45, 2019.
[9] Z.-H. Zhou, “A brief introduction to weakly supervised learning,” National Science Review, vol. 5, no. 1, pp. 44–53, 2017.
[10] D.-H. Lee, “Pseudo-label: The simple and efficient semi-supervised learning method for deep neural networks,” In ICML Workshop, 2013.
[11] H. Inoue, “Data augmentation by pairing samples for images classification,” arXiv preprint arXiv:1801.02929, 2018.
[12] I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y. Bengio, “Generative adversarial nets,” In NIPS, 2014.
[13] C. Szegedy, Wei Liu, Yangqing Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke, and A. Rabinovich, “Going deeper with convolutions,” 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015.
[14] S. Liu and W. Deng, “Very deep convolutional neural network based image classification using small training sample size,” 2015 3rd IAPR Asian Conference on Pattern Recognition (ACPR), 2015.
[15] K. He, X. Zhang, S. Ren, and J. Sun, “Deep Residual Learning for Image Recognition,” 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016.
[16] G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinberger, “Densely Connected Convolutional Networks,” 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017.
[17] T.-Y. Lin, P. Goyal, R. Girshick, K. He, and P. Dollar, “Focal Loss for Dense Object Detection,” 2017 IEEE International Conference on Computer Vision (ICCV), 2017.
[18] F. Milletari, N. Navab, and S.-A. Ahmadi, “V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation,” 2016 Fourth International Conference on 3D Vision (3DV), 2016.
[19] S. S. Salehi, D. Erdogmus, and A. Gholipour, “Tversky Loss Function for Image Segmentation Using 3D Fully Convolutional Deep Networks,” Machine Learning in Medical Imaging, pp. 379–387, 2017.
[20] B. Settles and M. Craven, “An analysis of active learning strategies for sequence labeling tasks,” Proceedings of the Conference on Empirical Methods in Natural Language Processing - EMNLP '08, 2008.
[21] W. Luo, A. Schwing, and R. Urtasun, “Latent structured active learning,” In Advances in Neural Information Processing Systems, pages 728–736, 2013.
[22] A. J. Joshi, F. Porikli, and N. Papanikolopoulos, “Multi-class active learning for image classification,” 2009 IEEE Conference on Computer Vision and Pattern Recognition, 2009.
[23] S. Chakraborty, V. Balasubramanian, Q. Sun, S. Panchanathan, and J. Ye, “Active Batch Selection via Convex Relaxations with Guaranteed Solution Bounds,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 37, no. 10, pp. 1945–1958, 2015.
[24] Z. Zhou, J. Shin, L. Zhang, S. Gurudu, M. Gotway, and J. Liang, “Fine-Tuning Convolutional Neural Networks for Biomedical Image Analysis: Actively and Incrementally,” 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017.
[25] Marc Gorriz, Axel Carlier, Emmanuel Faure, and Xavier Giro-i Nieto, “Cost-effective active learning for melanoma segmentation,” arXiv preprint arXiv:1711.09168, 2017.
[26] M. Lubrano di Scandalea, C. S Perone, M. Boudreau, and J. Cohen-Adad, “Deep active learning for axon-myelin segmentation on histology data,” CoRR, abs/1907.05143, 2019.
[27] Firat Ozdemir, Zixuan Peng, Philipp Fuernstahl, Christine Tanner, and Orcun Goksel, “Active learning for segmentation based on bayesian sample queries,” arXiv preprint arXiv:1912.10493, 2019.
[28] L. Yang, Y. Zhang, J. Chen, S. Zhang, and D. Z. Chen, “ Suggestive annotation: A deep active learning framework for biomedical image segmentation,” arXiv preprint arXiv:1706.04737, 2017.
[29] Y. Siddiqui, J. Valentin, and M. Niessner, “ViewAL: Active Learning With Viewpoint Entropy for Semantic Segmentation,” 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020.
[30] O. Oktay et al., “Attention U-Net: Learning where to look for the pancreas,” arXiv:1804.03999, 2018.
[31] Ming Liang and Xiaolin Hu, “Recurrent convolutional neural network for object recognition,” 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015.
[32] X. Glorot, A. Bordes, and Y. Bengio, “Deep sparse rectifier networks,” In Proceedings of the 14th International Conference on Artificial Intelligence and Statistics, pages 315–323,
2011.
[33] V. Nair and G. E. Hinton, “Rectified linear units improve restricted boltzmann machines,” In ICML, 2010.
[34] J. Bergstra, G. Desjardins, P. Lamblin, and Y. Bengio, “Quadratic polynomials learn better image features,” Technical report, Technical Report 1337, Departement d’Informatiqueet de Recherche Operationnelle, Universit´e de Montr´eal, 2009.
2009.
[35] X. Glorot and Y. Bengio, “Understanding the difficulty of training deep feedforward neural networks,” in Proc. AISTATS, 2010, pp. 249–256.
[36] D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization” In International Conference on Learning Representations (ICLR), 2015.
[37] J. Duchi, E. Hazan, and Y. Singer, “Adaptive subgradient methods for online learning and stochastic optimization,” Journal of Machine Learning Research, vol. 12, no. Jul, pp. 2121–2159, 2011.
[38] Tieleman, T. and Hinton, G. Lecture 6.5 - RMSProp, COURSERA: Neural Networks for Machine Learning. Technical report, 2012.
[39] W. R. Crum, O. Camara, and D. L. G. Hill, “Generalized Overlap Measures for Evaluation and Validation in Medical Image Analysis,” IEEE Transactions on Medical Imaging, vol. 25, no. 11, pp. 1451–1461, 2006.
[40] H. Zhao, J. Shi, X. Qi, X. Wang, and J. Jia, “Pyramid Scene Parsing Network,” 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017.
[41] M. Yang, K. Yu, C. Zhang, Z. Li, and K. Yang, “DenseASPP for Semantic Segmentation in Street Scenes,” 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018.
[42] Y. Gal, R. Islam, and Z. Ghahramani, “Deep bayesian active learning with image data,” International Conference on Machine Learning (ICML), 2017.