簡易檢索 / 詳目顯示
| 研究生： |
莊竣淳 Chuang, Chuew-Chuen |
|---|---|
| 論文名稱： |
卷積類神經網路於胸部X光影像結核病偵測之研究 Tuberculosis Detection Using Convolutional Neural Network in Chest Radiographs |
| 指導教授： |
孫永年
Sun, Yung-Nien |
| 學位類別： |
碩士 Master |
| 系所名稱： |
電機資訊學院 - 資訊工程學系 Department of Computer Science and Information Engineering |
| 論文出版年： | 2019 |
| 畢業學年度： | 107 |
| 語文別： | 英文 |
| 論文頁數： | 59 |
| 中文關鍵詞： | X光影像 、分類 、偵測 、肺結核 、卷積類神經網路 |
| 外文關鍵詞： | radiograph image, classification, detection, Tuberculosis, Convolutional Neural Network |
| 相關次數： | 點閱：57 下載：0 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
結核病是全球十大致命疾病之一，根據2018年世界衛生組織的數據顯示[1]，全球有超過一千萬新發病例，150萬例死亡，而這些死亡的病例有80%以上來自開發中國家[2]，主要原因是這些國家的資源匱乏以及醫療設施不足，導致難以治療與預防。
為了有效治療病患與降低疾病傳播風險，需要一個快速且精準的診斷方式。主要診斷結核病的方法有兩種:1)醫師使用胸部X光影像進行診斷，其診斷時間短；2)結核菌培養試驗，其診斷時間長(約4-6週)。因此在早期結核病快速檢測，胸部X光影像診斷被視為首選。然而，胸部X光影像的診斷可能因人而異，由於結核病在胸部X光影像上的特徵不固定且難以觀察，導致時常會出現觀察者判斷不一致的情形發生。本論文即為針對此問題設計一胸部Ｘ光結核病偵測系統，以輔助醫師進行更精準的判斷
本論文中提出一個使用卷積類神經網路的結核病檢測系統，用於輔助醫師在胸部X光影像上的判讀。首先是收集病人的胸部X光影像，系統會透過該影像進行辨識與分類，得出該影像是否為結核病，若得出結果為陽性，則系統會進行病灶位置的偵測，圈出可能為病灶的位置以提供醫師參考；反之，若得出結果為陰性，則屬於非結核病。
本系統主要分為兩個部分，判斷是否為結核病影像和偵測結核病灶的位置。首先是判斷是否為結核病影像的部分，透過深度學習從已知結果的胸部X光影像集來訓練模型，使模型能夠學習出結核病與非結核病的特徵差異，藉此來判斷是否為結核病胸部X光影像，在這部分即可快速過濾非病患影像以達快篩效果。接著是檢測結核病灶位置，該部分同樣使用了深度學習的方法，我們會先將已知病灶位置的胸部X光影像提供給模型做訓練，訓練完畢的模型即可透過胸部X光中的特徵來做判讀，並且提供病灶位置以提供醫師參考，透過這個功能，我們希望能提高醫師對於胸部X光影像診斷的精準度以減少觀察者不一致問題。
此系統使用Python以及Tensorflow框架來實現。我們的方法在結核病影像的分類上表現出優異的準確性，在分類準確度上可以高達91.07%。在結核病灶的偵測方面，儘管結核病灶特徵非常複雜難以分辨，召回率也能達到84%，藉此希望可以有效的輔助醫師與醫檢師快速診斷。我們的方法與結果顯示了深度學習在結核病檢驗中，能夠提供一個更精準且快速的檢測方法。
Tuberculosis is one of the top ten deadly diseases in the world. According to the report of World Health Organization (WHO) in 2018 [1], there are more than 10 million cases developed worldwide, and 1.5 million deaths; in which more than 80% of these deaths are from developing countries [2]. The lack of resources and inadequate medical facilities are main reasons that makes it difficult to treat and prevent this disease in these countries.
In order to effectively treat patients and reduce the risk of disease transmission, a fast and accurate diagnosis is needed. There are two main methods for diagnosing tuberculosis: 1) The doctor uses a chest X-ray image for diagnosis; 2) Patients take tuberculosis culture test, but that has a long time (about 4-6 weeks) to get results. Therefore, chest X-ray imaging diagnosis is preferred in the early detection of tuberculosis. However, the diagnosis of chest X-ray images may be variant from different observers, and the characteristics of tuberculosis on chest X-ray images are not consistent and difficult to observe, resulting in difficulty of diagnosis and observer variation. Our target is to solve these problems to help doctors make more accurate diagnosis.
We propose a tuberculosis detection system using a convolutional neural network to assist doctors in the interpretation of chest X-ray images. First, the chest X-ray images of tuberculosis and non-tuberculosis patients are collected. The system will classify the images into tuberculosis positive or negative. If the result is positive, the system will further detect the location of the lesion which can be provided as a more precise reference for doctors.
The system is mainly divided into two parts, one is tuberculosis classification and the other is lesion detection. First, a deep learning convolutional neural network is trained from a labeled chest X-ray image set, so that the network can learn the difference between the tuberculosis and not tuberculosis images and finally has the ability to make a classification on them. In the second part, a convolutional neural network is also applied. The lesion positions in the chest X-ray images are provided for the training of the detection network. An accurate classification can give fast diagnosis, so as to reduce labor costs. By providing the location information, the accuracy of the doctor's diagnosis of chest X-ray images can be improved and observer variations can be reduced.
The system is developed by using Python with Tensorflow framework. The classification performance is remarkably high of 91.07% accuracy. For the detection, thought the identification of lesions are very complicated, the recall of our system can reach 84.62%. It can effectively help doctors and medical examiners to fast diagnose. The results show that our proposed method can provide a more accurate and fast detection of tuberculosis in chest X-ray images.
[1] "Global tuberculosis report 2018," World Health Organization (WHO), 2018.
[2] Wikipedia, "Tuberculosis," 2018. [Online]. Available: https://en.wikipedia.org/wiki/Tuberculosis.
[3] S. Jaeger, A. Karargyris, S. Antani and G. Thoma, "Detecting tuberculosis in radiographs using combined lung masks," Proc. Int. IEEE Eng. Med. Biol. Soc., pp. 4978-4981, 2012.
[4] I. Gabriella, K. S. A. and S. A. W., "Early Detection of Tuberculosis using Chest X-Ray (CXR) with Computer-Aided Diagnosis," 2nd International Conference on Biomedical Engineering (IBIOMED), pp. 76-79, 2018.
[5] L. G. C. Evalgelista and E. B. Guedes, "Computer-Aided Tuberculosis Detection from Chest X-Ray Images with Convolutional Neural Networks," XV Encontro Nacional de Inteligência Artificial e Computacional, pp. 518-527, 2018.
[6] E. J. Hwang, S. Park and K.-N. Jin, "Development and Validation of a Deep Learning–based Automatic Detection Algorithm for Active Pulmonary Tuberculosis on Chest Radiographs," Clinical and Technical Considerations. Clin. Infect. Dis., 2018.
[7] Z. Xue, S. Jaeger, S. Antani, L. R. Long, A. Karargyris, J. Siegelman, L. R. Folio and G. R. Thoma, "Localizing tuberculosis in chest radiographs with deep learning," Medical imaging 2018: imaging informatics for healthcare, research, and applications. International Society for Optics and Photonics, p. 105790U, 2018.
[8] J. Gu, Z. Wang, J. Kuen, L. Ma, A. Shahroudy and B. Shuai, "Recent advances in convolutional neural networks," arXiv preprint arXiv:1512.07108, 2015.
[9] D. H. Hubel and T. N. Wiesel, "Receptive fields of single neurones in the cat's striate cortex," J. Physiol. 148, pp. 574-591, 1959.
[10] K. Fukushima, "Neocognitron: A Self-organizing Neural Network Model for a Mechanism of Pattern Recognition Unaffected by Shift in Position," Biol. Cybernetics 36, pp. 193-202, 1980.
[11] Y. LeCun, L. Bottou, Y. Bengio and P. Haffner, "Gradient-based learning applied to document recognition.," Proc. IEEE 86, p. 2278–2324, 1998.
[12] I. S. G. E. H. Alex Krizhevsky, "ImageNet Classification with Deep Convolutional", In Proceedings of Neural Information Processing Systems, pp. 1097-1105, 2012.
[13] R. F. Matthew D. Zeiler, "Visualizing and Understanding Convolutional Networks," European Conference on Computer Vision, pp. 818-833, 2014.
[14] K. Simonyan and A. Zisserman, "Very Deep Convolutional Networks for Large-Scale Image Recognition," arXiv preprint arXiv:1409.1556, 2015.
[15] C. Szegedy, W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke and A. Rabinovich, "Going deeper with convolutions.," IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1-9, 2015.
[16] K. He, X. Zhang, S. Ren and J. Sun, "Deep Residual Learning for Image Recognition," IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. pp770-778, 2016.
[17] G. Huang, Z. Liu, L. v. d. Maaten and K. Q. Weinberger, "Densely Connected Convolutional Networks," The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 4700-4708, 2017.
[18] Y. James, "[資料分析&機器學習] 第5.1講: 卷積神經網絡介紹(Convolutional Neural Network)," 2017. [Online]. Available: https://medium.com/jameslearningnote/%E8%B3%87%E6%96%99%E5%88%86%E6%9E%90-%E6%A9%9F%E5%99%A8%E5%AD%B8%E7%BF%92-%E7%AC%AC5-1%E8%AC%9B-%E5%8D%B7%E7%A9%8D%E7%A5%9E%E7%B6%93%E7%B6%B2%E7%B5%A1%E4%BB%8B%E7%B4%B9-convolutional-neural-network-4f8249d65d4f.
[19] Opengate, "深度學習：使用激勵函數的目的、如何選擇激勵函數 Deep Learning : the role of the activation function," 26 02 2017. [Online]. Available: https://mropengate.blogspot.com/2017/02/deep-learning-role-of-activation.html.
[20] Wikipedia, "Rectifier (neural networks)," 2019. [Online]. Available: https://en.wikipedia.org/wiki/Rectifier_(neural_networks).
[21] S. Ioffe and C. Szegedy, "Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift," arXiv preprint arXiv:1502.03167, 2015.
[22] Wikipedia, "Batch normalization," 2019. [Online]. Available: https://en.wikipedia.org/wiki/Batch_normalization.
[23] Rumelhart, D. E., G. E. Hinton and R. J. Williams., "Learning representations by back-propagating errors," Cognitive modeling, pp. pp. 696-699 , 1988.
[24] D. P. Kingma and J. Ba, "Adam: A Method for Stochastic Optimization," arXiv preprint arXiv:1412.6980, 2014.
[25] Bushaev and Vitaly, "Adam — latest trends in deep learning optimization.," 22 10 2018. [Online]. Available: https://towardsdatascience.com/adam-latest-trends-in-deep-learning-optimization-6be9a291375c.
[26] F. Yu and V. Koltun, "Multi-Scale Context Aggregation by Dilated Convolutions," arXiv preprint arXiv:1511.07122, 2016.
[27] Darrell, F. Yu, D. Wang, E. Shelhamer and T. Darrell, "Deep Layer Aggregation," The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2403-2412, 2018.
[28] J. Hu, L. Shen, S. Albanie, G. Sun and E. Wu, "Squeeze-and-Excitation Networks," IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 7132-7141, 2018.
[29] H. Larochelle and G. Hinton, "Learning to combine foveal glimpses with a thirdorder boltzmann machine.," International Conference on Neural Information Processing Systems (NIPS), pp. 1243-1251, 2010.
[30] F. Wang, M. Jiang, C. Qian, S. Yang, C. Li, H. Zhang and X. Wang, "Residual attention network for image classification.," arXiv preprint arXiv:1704.06904, 2017.
[31] J. P. J.-Y. L. I. S. K. Sanghyun Woo, "CBAM: Convolutional Block Attention Module," in in The European Conference on Computer Vision (ECCV), pp. 3-19, 2018.
[32] S. Ren, K. He, R. Girshick and J. Sun, "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks," Pattern Analysis and Machine Intelligence, pp. 1137 - 1149, 2017.
[33] R. Girshick, J. Donahue, T. Darrell and J. Malik, "Rich feature hierarchies for accurate object detection and semantic segmentation," in IEEE Conference on Computer Vision and Pattern Recognition, pp. 580-587, 2014.
[34] J. Uijlings, K. v. d. Sande, T. Gevers and A. Smeulders, "Selective Search for Object Recognition," International Journal of Computer Vision, pp. pp. 154-171, 09 2013.
[35] T. Grel, "Region of interest pooling explained," 28 02 2017. [Online]. Available: https://deepsense.ai/region-of-interest-pooling-explained/.
[36] T.-Y. Lin, P. Dollár, R. Girshick, K. He, B. Hariharan and S. Belongie, "Feature pyramid networks for object detection.," IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2117-2125, 2017.
[37] Z.-X. Li and F.-Q. Zhou, "FSSD: Feature Fusion Single Shot Multibox Detector," arXiv preprint arXiv:1712.00960, 2018.
[38] P. D. R. G. Tsung-Yi Lin, K. He, B. Hariharan and S. Belongie, "Feature Pyramid Networks for Object Detection," in The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2117-2125, 2017.
[39] S. Liu, L. Qi, H. Qin, J. Shi and J. Jia, "Path Aggregation Network for Instance Segmentation," in The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.8759-8768, 2017.
[40] C. Liu, Y. Cao, M. Alcantara, B. Liu, M. Brunette, J. Peinado and W. Curioso, "TX-CNN: Detecting tuberculosis in chest X-ray images using convolutional neural network," IEEE International Conference on Image Processing (ICIP), pp. 2314-2318, 2017.
校內：2024-08-30公開校外：不公開