西元 2019 年全球爆發 2019 新型冠狀病毒 (COVID­19) 感染人數超過一千萬人，無疑 是近十年最嚴重的全球性傳染疾病。COVID­19 的爆發造成醫院出現待診人數激增， 而由於檢測方式的時間過長導致無法克服大量待診人數。CT 影像檢測是 COVID­19 檢測方式中方便性最高的一種，但由於是透過醫師的判斷來進行，因此容易出現診 斷精度隨醫師疲勞度下降的問題。醫師撰寫診斷報告需要花費大量時間標示病症部 位以及撰寫診斷細節，導致診斷效率低，導致傳染時間延長使得疾病擴張速度激增。
本篇針對此現狀提出一套基於深度學習的診斷系統 Deep Learning COVID19 Inspection System (COVID­IS)，透過深度學習自主學習的能力學習看診與報告撰寫來克服上述 提到的問題。COVID­IS 透過鎖定目標器官的方式，使得病症診斷不受到其他器官影 響，藉此降低資料差異性。本篇提出一套自適應性背景去除模型，在無需花費額外 的資料標籤處理時間進行訓練，使得背景去除模型具備消去不必要背景資訊提升辨 識準確度。系統透過上述兩機制使得在資料差異大的醫療 CT 圖片中具備良好應用 性，透過提升肺炎辨識模型的準確度達到自動判斷病症區域再由醫師進行確認，藉 此降低醫師看診的主觀性並且降低診斷精度隨時間下降的發生率。COVID­IS 系統病 症輪廓標示上透過本篇提出的語意嵌入式 U­Net，根據系統辨識出來的病症框下進行 精確的病症輪廓切割，降低醫師的診斷報告撰寫時間。
本篇利用空開資料 COVID­CT 驗證 Deep Learning COVID19 Inspection System (COVID­ IS) 的有效性。驗證結果證實 COVID­IS 鎖定器官的準確性，以及透過過濾不重要背 景資訊將辨識率提升 16 %。系統最終也提供疾病輪廓標籤圖，驗證系統確實達到自 主診斷以及輔助撰寫報告的能力。

The COVID­19 is the most serious global infectious disease in the past decade. If the pa­tient didn’t have the travel history in six month, the hospital will use CT image to check. But this way has high subjective and need long time to write report. So our proposed Deep Learning COVID­19 Inspection System (COVID­IS) based on deep learning for automated recognition to improve efficiency and reduce subjective. The case studies in open data COVID­CT show the COVID­IS uses the adaptability background removing model to improve 17% detection sensitivity, and uses the SEU-­Net to do precisely dis­ ease outline segmentation.

[1] S. Khan and S.­P. Yong, “A deep learning architecture for classifying medical images of anatomy object,” in Asia­Pacific Signal and Information Processing Association An­ nual Summit and Conference), 2017.
[2] Q. Liu, L. Yu, L. Luo, Q. Dou, and P. A. Heng, “Semi­supervised medical image clas­ sification with relation­driven self­ensembling model,” 2020.
[3]O.Emad,I.A.Yassine,andA.S.Fahmy,“Automaticlocalizationoftheleftventriclein cardiac mri images using deep learning,” in 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2015, pp. 683–686.
[4] M. A. Ghamdi, M. Abdel­Mottaleb, and F. Collado­Mesa, “Du­net: Convolutional net­ work for the detection of arterial calcifications in mammograms,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[5] X. Wang, X. Deng, Q. Fu, Q. Zhou, J. Feng, H. Ma, W. Liu, and C. Zheng, “A weakly­ supervised framework for covid­19 classification and lesion localization from chest ct,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[6] N. Wu, J. Phang, J. Park, Y. Shen, Z. Huang, M. Zorin, S. Jastrzębski, T. Févry, J. Kat­ snelson, E. Kim, S. Wolfson, U. Parikh, S. Gaddam, L. L. Y. Lin, K. Ho, J. D. Wein­ stein, B. Reig, Y. Gao, H. Toth, K. Pysarenko, A. Lewin, J. Lee, K. Airola, E. Mema, S. Chung, E. Hwang, N. Samreen, S. G. Kim, L. Heacock, L. Moy, K. Cho, and K. J. Geras, “Deep neural networks improve radiologists’ performance in breast cancer screening,” 2019.
[7] O. Ronneberger, P. Fischer, and T. Brox, “U­net: Convolutional networks for biomed­ ical image segmentation,” 2015.
[8] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” 2015.
[9] J. Redmon and A. Farhadi, “Yolov3: An incremental improvement,” 2018.
[10] J. Youn, M. L. Ommen, M. B. Stuart, E. V. Thomsen, N. B. Larsen, and J. A. Jensen, “Detection and localization of ultrasound scatterers using convolutional neural net­ works,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[11] S. Roy, W. Menapace, S. Oei, B. Luijten, E. Fini, C. Saltori, I. Huijben, N. Chen­ nakeshava, F. Mento, A. Sentelli, E. Peschiera, R. Trevisan, G. Maschietto, E. Torri, R. Inchingolo, A. Smargiassi, G. Soldati, P. Rota, A. Passerini, R. J. G. Van Sloun, E. Ricci, and L. Demi, “Deep learning for classification and localization of covid­19 markers in point­of­care lung ultrasound,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[12] S. Liang, K. Thung, D. Nie, Y. Zhang, and D. Shen, “Multi­view spatial aggregation framework for joint localization and segmentation of organs at risk in head and neck ct images,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[13] J. M. H. Noothout, B. D. De Vos, J. M. Wolterink, E. M. Postma, P. A. M. Smeets, R. A. P. Takx, T. Leiner, M. A. Viergever, and I. Išgum, “Deep learning­based regres­ sion and classification for automatic landmark localization in medical images,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[14] J. Torrents­Barrena, G. Piella, E. Gratacós, E. Eixarch, M. Ceresa, and M. A. G. Ballester, “Deep q­capsnet reinforcement learning framework for intrauterine cavity segmentation in ttts fetal surgery planning,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[15] K. Murphy, B. Ginneken, S. Klein, M. Staring, B. Hoop, M. Viergever, and J. Pluim, “Semi­automatic construction of reference standards for evaluation of image registra­ tion,” Medical image analysis, vol. 15, pp. 71–84, 02 2011.
[16] O. Oktay, W. Bai, R. Guerrero, M. Rajchl, A. de Marvao, D. P. O'Regan, S. A. Cook, M. P. Heinrich, B. Glocker, and D. Rueckert, “Stratified decision forests for accurate anatomical landmark localization in cardiac images,” IEEE Transactions on Medical Imaging, vol. 36, no. 1, pp. 332–342, 2017.
[17] M. Buscema, “Back propagation neural networks,” Substance use misuse, vol. 33, pp. 233–70, 02 1998.
[18] W. A. Al, H. Y. Jung, I. D. Yun, Y. Jang, H.­B. Park, and H.­J. Chang, “Automatic aortic valve landmark localization in coronary ct angiography using colonial walk,” PLOS ONE, vol. 13, no. 7, pp. 1–23, 07 2018. [Online]. Available: https://doi.org/10.1371/journal.pone.0200317
[19] N. Torosdagli, D. K. Liberton, P. Verma, M. Sincan, J. S. Lee, and U. Bagci, “Deep geodesic learning for segmentation and anatomical landmarking,” IEEE Transactions on Medical Imaging, vol. 38, no. 4, pp. 919–931, 2019.
[20] D. Perkins and G. Salomon, “Transfer of learning,” vol. 11, 07 1999.
[21] B. Subedi, “Emerging trends of research on transfer of learning,” International Education Journal, vol. 5, 01 2004.
[22] I. Isgum, M. Staring, A. Bartels­Rutten, M. Prokop, M. Viergever, and B. Ginneken, “Multi­atlas­based segmentation with local decision fusion—application to cardiac and aortic segmentation in ct scans,” Medical Imaging, IEEE Transactions on, vol. 28, pp. 1000 – 1010, 08 2009.
[23] D. Štern, T. Ebner, and M. Urschler, “From local to global random regression forests: Exploring anatomical landmark localization,” 10 2016.
[24] C. Lindner and T. Cootes, “Fully automatic cephalometric evaluation using random forest regression­voting,” 04 2015.
[25] G. Litjens, T. Kooi, B. E. Bejnordi, A. A. A. Setio, F. Ciompi, M. Ghafoorian, J. A. van der Laak, B. van Ginneken, and C. I. Sánchez, “A survey on deep learning in medical image analysis,” Medical Image Analysis, vol. 42, pp. 60 – 88, 2017. [Online]. Available: http://www.sciencedirect.com/science/article/pii/S1361841517301135
[26] Y. Zheng, D. Liu, B. Georgescu, H. Nguyen, and D. Comaniciu, “3d deep learning for efficient and robust landmark detection in volumetric data,” in Medical Image Com­ puting and Computer­Assisted Intervention – MICCAI 2015, N. Navab, J. Hornegger, W. M. Wells, and A. Frangi, Eds. Cham: Springer International Publishing, 2015, pp. 565–572.
[27] D. Yang, S. Zhang, Z. Yan, C. Tan, K. Li, and D. Metaxas, “Automated anatomical landmark detection ondistal femur surface using convolutional neural network,” in 2015 IEEE 12th International Symposium on Biomedical Imaging (ISBI), 2015, pp. 17–21.
[28] C. Payer, D. Štern, H. Bischof, and M. Urschler, “Integrating spatial configuration into heatmap regression based cnns for landmark localization,” Medical Image Analysis, vol. 54, pp. 207 – 219, 2019. [Online]. Available: http://www.sciencedirect.com/ science/article/pii/S1361841518305784
[29] J. Zhang, M. Liu, and D. Shen, “Detecting anatomical landmarks from limited medical imaging data using two­stage task­oriented deep neural networks,” IEEE Transactions on Image Processing, vol. 26, no. 10, pp. 4753–4764, 2017.
[30] Y. Li, A. Alansary, J. J. Cerrolaza, B. Khanal, M. Sinclair, J. Matthew, C. Gupta, C. Knight, B. Kainz, and D. Rueckert, “Fast multiple landmark localisation using a patch­based iterative network,” 2018.
[31] F. Ghesu, B. Georgescu, Y. Zheng, S. Grbic, A. Maier, J. Hornegger, and D. Comani­ ciu, “Multi­scale deep reinforcement learning for real­time 3d­landmark detection in ct scans,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 41, no. 1, pp. 176–189, 2019.
[32] A. Alansary, O. Oktay, Y. Li, L. L. Folgoc, B. Hou, G. Vaillant, K. Kamnitsas, A. Vlontzos, B. Glocker, B. Kainz, and D. Rueckert, “Evaluating reinforcement learning agents for anatomical landmark detection,” Medical Image Analysis, vol. 53, pp. 156 – 164, 2019. [Online]. Available: http://www.sciencedirect.com/science/ article/pii/S1361841518306121
[33] L. Zhou, Z. Li, J. Zhou, H. Li, Y. Chen, Y. Huang, D. Xie, L. Zhao, M. Fan, S. Hashmi, F. AbdelKareem, R. Eiada, X. Xiao, L. Li, Z. Qiu, and X. Gao, “A rapid, accurate and machine­agnostic segmentation and quantification method for ct­based covid­19 diagnosis,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[34] G.Wang,X.Liu,C.Li,Z.Xu,J.Ruan,H.Zhu,T.Meng,K.Li,N.Huang,andS.Zhang, “A noise­robust framework for automatic segmentation of covid­19 pneumonia lesions from ct images,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[35] D. Fan, T. Zhou, G. Ji, Y. Zhou, G. Chen, H. Fu, J. Shen, and L. Shao, “Inf­net: Au­ tomatic covid­19 lung infection segmentation from ct images,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[36] Z. Han, B. Wei, Y. Hong, T. Li, J. Cong, X. Zhu, H. Wei, and W. Zhang, “Accurate screening of covid­19 using attention based deep 3d multiple instance learning,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[37] X. Ouyang, J. Huo, L. Xia, F. Shan, J. Liu, Z. Mo, F. Yan, Z. Ding, Q. Yang, B. Song, F. Shi, H. Yuan, Y. Wei, X. Cao, Y. Gao, D. Wu, Q. Wang, and D. Shen, “Dual­sampling attention network for diagnosis of covid­19 from community acquired pneumonia,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[38] W. Xie, C. Jacobs, J. Charbonnier, and B. van Ginneken, “Relational modeling for robust and efficient pulmonary lobe segmentation in ct scans,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[39] Y. Oh, S. Park, and J. C. Ye, “Deep learning covid­19 features on cxr using limited training data sets,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[40] H. Kang, L. Xia, F. Yan, Z. Wan, F. Shi, H. Yuan, H. Jiang, D. Wu, H. Sui, C. Zhang, and D. Shen, “Diagnosis of coronavirus disease 2019 (covid­19) with structured latent multi­view representation learning,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[41] S. Karanam, R. Li, F. Yang, W. Hu, T. Chen, and Z. Wu, “Towards contactless patient positioning,” IEEE Transactions on Medical Imaging, pp. 1–1, 2020.
[42] H. Shin, H. R. Roth, M. Gao, L. Lu, Z. Xu, I. Nogues, J. Yao, D. Mollura, and R. M. Summers, “Deep convolutional neural networks for computer­aided detection: Cnn ar­ chitectures, dataset characteristics and transfer learning,” IEEE Transactions on Medi­ cal Imaging, vol. 35, no. 5, pp. 1285–1298, 2016.
[43] N. Wu, J. Phang, J. Park, Y. Shen, Z. Huang, M. Zorin, S. Jastrzębski, T. Févry, J. Kat­ snelson, E. Kim, S. Wolfson, U. Parikh, S. Gaddam, L. L. Y. Lin, K. Ho, J. D. Wein­ stein, B. Reig, Y. Gao, H. Toth, K. Pysarenko, A. Lewin, J. Lee, K. Airola, E. Mema, S. Chung, E. Hwang, N. Samreen, S. G. Kim, L. Heacock, L. Moy, K. Cho, and K. J. Geras, “Deep neural networks improve radiologists'performance in breast can­ cer screening,” IEEE Transactions on Medical Imaging, vol. 39, no. 4, pp. 1184–1194, 2020.
[44] J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic seg­ mentation,” in 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015, pp. 3431–3440.
[45] M.Gao,Z.Xu,L.Lu,A.Wu,I.Nogues,R.M.Summers,andD.J.Mollura,“Segmenta­ tion label propagation using deep convolutional neural networks and dense conditional random field,” in 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI), 2016, pp. 1265–1268.
[46] X. Zhu, C. Vondrick, D. Ramanan, and C. Fowlkes, “Do we need more training data or better models for object detection?” BMVC 2012 ­ Electronic Proceedings of the British Machine Vision Conference 2012, vol. 3, 01 2012.
[47] Y. Jia, E. Shelhamer, J. Donahue, S. Karayev, J. Long, R. Girshick, S. Guadarrama, and T. Darrell, “Caffe: Convolutional architecture for fast feature embedding,”in Proceedings of the 22Nd ACM International Conference on Multimedia, ser. MM ’14. New York, NY, USA: ACM, 2014, pp. 675–678. [Online]. Available: http://doi.acm.org/10.1145/2647868.2654889
[48]T.KooiandN.Karssemeijer,“Classifyingsymmetricaldifferencesandtemporalchange in mammography using deep neural networks,” 2017.
[49] I. D. Apostolopoulos and T. Bessiana, “Covid­19: Automatic detection from x­ray im­ ages utilizing transfer learning with convolutional neural networks,” 2020.
[50]I.D.Apostolopoulos,S.Aznaouridis,andM.Tzani,“Extractingpossiblyrepresentative covid­19 biomarkers from x­ray images with deep learning approach and image data related to pulmonary diseases,” 2020.
[51] S. Salehi, A. Abedi, S. Balakrishnan, and A. Gholamrezanezhad, “Coronavirus disease 2019 (covid­19): A systematic review of imaging findings in 919 patients,” American Journal of Roentgenology, vol. 215, pp. 1–7, 03 2020.
[52] S. H. Yoon, K. Lee, J. Kim, Y. Lee, H. Ko, K. Kim, C. M. Park, and Y.­H. Kim, “Chest radiographic and ct findings of the 2019 novel coronavirus disease (covid­19): Analysis of nine patients treated in korea,” Korean Journal of Radiology, vol. 21, 02 2020.
[53] M. Z. Alom, M. M. S. Rahman, M. S. Nasrin, T. M. Taha, and V. K. Asari, “Covid_mtnet: Covid19 detection with multitask deep learning approaches,” 2020.
[54] O. Elharrouss, N. Subramanian, and S. Al­Maadeed, “An encoder­decoder­based method for covid19 lung infection segmentation,” 2020.
[55] X. Yang, X. He, J. Zhao, Y. Zhang, S. Zhang, and P. Xie, “Covid­ct­dataset: A ct scan dataset about covid­19,” 2020.
[56]S.H.Kang,K.Jeon,H.­J.Kim,J.K.Seo,andS.­H.Lee,“Automaticthree­dimensional cephalometric annotation system using three­dimensional convolutional neural net­ works,” 2018.
[57] S. Gupta, R. Girshick, P. Arbeláez, and J. Malik, “Learning rich features from rgb­d images for object detection and segmentation,” 2014.
[58]L.­C.Chen,G.Papandreou,I.Kokkinos,K.Murphy,andA.L.Yuille,“Semanticimage segmentation with deep convolutional nets and fully connected crfs,” 2014.
[59] W.Lotter,G.Sorensen,andD.Cox,“Amulti­scalecnnandcurriculumlearningstrategy for mammogram classification,” 2017.
[60] E.E.­D.Hemdan,M.A.Shouman,andM.E.Karar,“Covidx­net:Aframeworkofdeep learning classifiers to diagnose covid­19 in x­ray images,” 2020.
[61] M. Farooq and A. Hafeez, “Covid­resnet: A deep learning framework for screening of covid19 from radiographs,” 2020.
[62] P. Afshar, S. Heidarian, F. Naderkhani, A. Oikonomou, K. N. Plataniotis, and A. Mo­ hammadi, “Covid­caps: A capsule network­based framework for identification of covid­19 cases from x­ray images,” 2020.