深度學習已經在醫學圖像被廣泛的應用，包括去噪、分類和分割等。這些應用都是為了在臨床評估過程中為放射科醫師提供更多的信息，從而提高評估的準確性。近年來，許多醫學去噪方法在定量和定性上都顯示出了顯著的偽影和噪聲的去除效果。然而，現有的方法都是針對於人眼視覺進行去噪，即它們的設計是為了盡量減少人眼所能感知的噪聲。在本文中，我們提出了一個應用導向的去噪框架，該框架專注於為其後的應用神經網絡進行去噪。在我們的實驗中，我們將提出的框架應用於不同的數據集、模型和任務。實驗結果表明，與人眼視覺去噪框架相比，該框架具有更好的去噪效果。

Deep learning had already demonstrated its power in medical images, including denoising, classification, segmentation, etc. All these applications are proposed to automatically analyze medical images beforehand, which brings more information to radiologists during clinical assessment for accuracy improvement. Recently, many medical denoising methods had shown their significant artifact reduction result and noise removal both quantitatively and qualitatively. However, those existing methods are developed around human-vision, i.e., they are designed to minimize the noise effect that can be perceived by human eyes. In this thesis, we introduce an application-guided denoising framework, which focuses on denoising for the following neural networks. In our experiments, we apply the proposed framework to different datasets, models, and use cases. Experimental results show that our proposed framework can achieve a better result than human-vision denoising network.

[1] X. Xu, Y. Ding, S. X. Hu, M. Niemier, J. Cong, Y. Hu, and Y. Shi, “Scaling for edgeinference of deep neural networks,”Nature Electronics, vol. 1, no. 4, pp. 216–222,2018.
[2] Y. Ding, J. Liu, X. Xu, M. Huang, J. Zhuang, J. Xiong, and Y. Shi, “Uncertainty-aware training of neural networks for selective medical image segmentation,” inMedi-cal Imaging with Deep Learning, 2020.
[3] X. Xu, Q. Lu, L. Yang, S. Hu, D. Chen, Y. Hu, and Y. Shi, “Quantization of fullyconvolutional networks for accurate biomedical image segmentation,” inCVPR, 2018,pp. 8300–8308.
[4] T. Wang, X. Xu, J. Xiong, Q. Jia, H. Yuan, M. Huang, J. Zhuang, and Y. Shi, “Ica-unet: Ica inspired statistical unet for real-time 3d cardiac cine mri segmentation,” inMICCAI, 2020, pp. 447–457.
[5] X. Xu, T. Wang, Y. Shi, H. Yuan, Q. Jia, M. Huang, and J. Zhuang, “Whole heart andgreat vessel segmentation in congenital heart disease using deep neural networks andgraph matching,” inMICCAI, 2019, pp. 477–485.35
[6] X. Xu, T. Wang, J. Zhuang, H. Yuan, M. Huang, J. Cen, Q. Jia, Y. Dong, and Y. Shi,“Imagechd: A 3d computed tomography image dataset for classification of congenitalheart disease,” inMICCAI, 2020, pp. 77–87.
[7] F. E. Boas, and D. Fleischmann, “Ct artifacts: causes and reduction techniques,”Imag-ing Med, vol. 4, no. 2, pp. 229–240, 2012.
[8] K. Krupa, and M. Bekiesi ́nska-Figatowska, “Artifacts in magnetic resonance imaging,”Polish journal of radiology, vol. 80, p. 93, 2015.[9] J. Liu, Y. Ding, J. Xiong, Q. Jia, M. Huang, J. Zhuang, B. Xie, C.-C. Liu, and Y. Shi,“Multi-cycle-consistent adversarial networks for ct image denoising,” inISBI, 2020,pp. 614–618.
[10] F. Isensee, P. Kickingereder, W. Wick, M. Bendszus, and K. H. Maier-Hein, “Nonew-net,” inInternational MICCAI Brainlesion Workshop, 2018, pp. 234-244.
[11] H. H. Sultan, N. M. Salem, and W. Al-Atabany, “Multi-classification of brain tumorimages using deep neural network,”IEEE Access, vol. 7, pp. 69215–69225, 2019.
[12] Y.-J. Chen, Y.-J. Chang, S.-C. Wen, Y. Shi, X. Xu, T.-Y. Ho, Q. Jia, M. Huang, and J.Zhuang, “Zero-Shot Medical Image Artifact Reduction,” inISBI, 2020, pp. 862–866.
[13] H. Chen , Y. Zhang, M. K. Kalra, F. Lin, Y. Chen, P. Liao, J. Zhou, and G. Wang,“Low-dose ct with a residual encoder-decoder convolutional neural network,”IEEEtransactions on medical imaging, vol. 36, no. 12, pp. 2524-2535, 2017.
[14] D. Jiang, W. Dou, L. Vosters, X. Xu, Y. Sun, and T. Tan, “Denoising of 3d mag-netic resonance images with multi-channel residual learning of convolutional neuralnetwork,”Japanese journal of radiology, vol. 36, no. 9, pp. 566-574, 2018.36
[15] D. Liu, B. Wen, J. Jiao, X. Liu, Z. Wang and T. S. Huang, “Connecting Image De-noising and High-Level Vision Tasks via Deep Learning,”IEEE Transactions on ImageProcessing, vol. 29, pp. 3695–3706, 2020.
[16] Z. Fan, L. Sun, X. Ding, Y. Huang, C. Cai, and J. Paisley, “A Segmentation-AwareDeep Fusion Network for Compressed Sensing MRI,” inECCV, 2018, pp. 55–70.
[17] C. Szegedy, W. Zaremba, I. Sutskever, J. Bruna, D. Erhan, I. Goodfellow, and R.Fergus, “Intriguing properties of neural networks,” arXiv preprint arXiv:1312.6199,2013.
[18] X. Zhuang, “Challenges and methodologies of fully automatic whole heart segmen-tation: a review,”Journal of healthcare engineering, vol. 4, no. 3, pp. 371-407, 2013.
[19] B.H. Menze, A. Jakab, S. Bauer, J. Kalpathy-Cramer, K. Farahani, J. Kirby, Y. Burren,N. Porz, J. Slotboom, R. Wiest, et al., “The multimodal brain tumor image segmenta-tion benchmark (brats),”IEEE transactions on medical imaging, vol. 34, no. 10, pp.1993-2024, 2014.[20] J. Cheng, “brain tumor dataset,” 2016.
[21] K. Zhang, W. Zuo, Y. Chen, D. Meng, and L. Zhang, “Beyond a gaussian denoiser:Residual learning of deep cnn for image denoising,”IEEE Transactions on Image Pro-cessing, vol. 26, no. 7, pp. 3142–3155, 2017.
[22] A. Mortazi, J. Burt, and U. Bagci, “Multi-planar deep segmentation networks forcardiac substructures from mri and ct,” inInternational Workshop on Statistical Atlasesand Computational Models of the Heart, 2017, pp. 199-206.37
[23] A. Zhao, G. Balakrishnan, F. Durand, J. V. Guttag, and A. V. Dalca, “Data aug-mentation using learned transformations for one-shot medical image segmentation,”inCVPR, 2019, pp. 8543-8553.[24] D. Karimi, and S. E. Salcudean, “Reducing the hausdorffdistance in medical imagesegmentation with convolutional neural networks,” arXiv preprint arXiv:1904.10030,2019.
[25] M. Arnold, I. Soerjomataram, J. Ferlay, and D. Forman, “Global incidence of oe-sophageal cancer by histological subtype in 2012,”Gut, vol. 64, no. 3, pp. 381–387,2014.
[26] K. M. Sjoquist, B. H. Burmeister, B. M. Smithers, J. R. Zalcberg, R. J. Simes, A. Bar-bour, and V. Gebski, “Survival after neoadjuvant chemotherapy or chemoradiotherapyfor resectable oesophageal carcinoma: an updated meta-analysis,”The Lancet Oncol-ogy, vol. 12, no. 7, pp. 681–692, 2011.
[27] V.H Pieter, H. MCCM, V.L JJB, S. EW, H. MI Van Berge, W. BPL, et al., “Preopera-tive chemoradiotherapy for esophageal or junctional cancer,”New England Journal ofMedicine, vol. 366, no. 22, pp. 2074-2084, 2012
[28] U. Ronellenfitsch, M. Schwarzbach, R. Hofheinz, P. Kienle, M. Kieser, T. E. Slanger,and K. Jensen, “Perioperative chemo(radio)therapy versus primary surgery for re-sectable adenocarcinoma of the stomach, gastroesophageal junction, and lower esoph-agus,”Cochrane Database of Systematic Reviews, 2013.
[29] A. H. H ̈olscher, E. Bollschweiler, D. Bogoevski, H. Schmidt, R. Semrau, and J. R.Izbicki, “Prognostic impact of neoadjuvant chemoradiation in cT3 oesophageal cancer38
– A propensity score matched analysis,”European Journal of Cancer, vol. 50, no. 17,pp. 2950–2957, 2014.
[30] P. M. Schneider, S. E. Baldus, R. Metzger, M. Kocher, R. Bongartz, E. Bollschweiler,H. Schaefer, J. Thiele, H. P. Dienes, R. P. Mueller, and A. H. Hoelscher, “Histomorpho-logic Tumor Regression and Lymph Node Metastases Determine Prognosis FollowingNeoadjuvant Radiochemotherapy for Esophageal Cancer,”Annals of Surgery, vol. 242,no. 5, pp. 684–692, 2005.
[31] E. Bollschweiler, R. Metzger, U. Drebber, S. Baldus, D. Vallb ̈ohmer, M. Kocher, andA. H ̈olscher, “Histological type of esophageal cancer might affect response to neo-adjuvant radiochemotherapy and subsequent prognosis,”Annals of Oncology, vol. 20,no. 2, pp. 231–238, 2009.
[32] A. C. Berger, J. Farma, W. J. Scott, G. Freedman, L. Weiner, J. D. Cheng, H.Wang, and M. Goldberg, “Complete Response to Neoadjuvant Chemoradiotherapy inEsophageal Carcinoma Is Associated With Significantly Improved Survival,”Journalof Clinical Oncology, vol. 23, no. 19, pp. 4330–4337, 2005.
[33] J. Tepper, M. J. Krasna, D. Niedzwiecki, D. Hollis, C. E. Reed, R. Goldberg, K. Kiel,C. Willett, D. Sugarbaker, and R. Mayer, “Phase III Trial of Trimodality Therapy WithCisplatin, Fluorouracil, Radiotherapy, and Surgery Compared With Surgery Alone forEsophageal Cancer: CALGB 9781,”Journal of Clinical Oncology, vol. 26, no. 7, pp.1086–1092, 2008.
[34] Y. Hamai, J. Hihara, M. Emi, Y. Murakami, M. Kenjo, Y. Nagata, and M. Okada,“Results of Neoadjuvant Chemoradiotherapy With Docetaxel and 5-Fluorouracil Fol-39
lowed by Esophagectomy to Treat Locally Advanced Esophageal Cancer,”The Annalsof Thoracic Surgery, vol. 99, no. 6, pp. 1887–1893, 2015.[35] L. Bedenne, P. Michel, O. Bouch ́e, C. Milan, C. Mariette, T. Conroy, D. Pezet,B. Roullet, J.-F. Seitz, J.-P. Herr, B. Paillot, P. Arveux, F. Bonnetain, and C. Bin-quet, “Chemoradiation Followed by Surgery Compared With Chemoradiation Alonein Squamous Cancer of the Esophagus: FFCD 9102,”Journal of Clinical Oncology,vol. 25, no. 10, pp. 1160–1168, 2007.
[36] K. Yamashita, N. Katada, H. Moriya, K. Hosoda, H. Mieno, C. Katada, W. Koizumi,K. Hoshi, and M. Watanabe, “Neoadjuvant chemotherapy of triplet regimens ofdocetaxel/cisplatin/5-FU (DCF NAC) may improve patient prognosis of cStage II/IIIesophageal squamous cell carcinoma-propensity score analysis,”General Thoracic andCardiovascular Surgery, vol. 64, no. 4, pp. 209–215, 2016.
[37] C. M. D. Bakker, J. K. Smit, A. M. E. Bruynzeel, N. C. T. V. Grieken, F. Daams,S. Derks, M. A. Cuesta, J. T. M. Plukker, and D. L. V. D. Peet, “Non responders toneoadjuvant chemoradiation for esophageal cancer: why better prediction is neces-sary,”Journal of Thoracic Disease, vol. 9, no. S8, 2017.
[38] Y. Li, H.-C. Huang, L.-Q. Chen, L.-Y. Xu, E.-M. Li, and J.-J. Zhang, “Predictivebiomarkers for response of esophageal cancer to chemo(radio)therapy: A systematicreview and meta-analysis,”Surgical Oncology, vol. 26, no. 4, pp. 460–472, 2017.
[39] C. Wong and S. Law, “Predictive factors in the evaluation of treatment response toneoadjuvant chemoradiotherapy in patients with advanced esophageal squamous cellcancer,”Journal of Thoracic Disease, vol. 9, no. S8, 2017.40
[40] F. Zhang, S. Yao, Z. Li, C. Liang, K. Zhao, Y. Huang, Y. Gao, J. Qu, Z. Li, and Z. Liu,“Predicting treatment response to neoadjuvant chemoradiotherapy in local advancedrectal cancer by biopsy digital pathology image features,”Clinical and TranslationalMedicine, vol. 10, no. 2, 2020.
[41] R. Turkki, D. Byckhov, M. Lundin, J. Isola, S. Nordling, P. E. Kovanen, C. Verrill,K. V. Smitten, H. Joensuu, J. Lundin, and N. Linder, “Breast cancer outcome predic-tion with tumour tissue images and machine learning,”Breast Cancer Research andTreatment, vol. 177, no. 1, pp. 41–52, 2019.
[42] O.-J. Skrede, S. D. Raedt, A. Kleppe, T. S. Hveem, K. Liestøl, J. Maddison, H. A.Askautrud, M. Pradhan, J. A. Nesheim, F. Albregtsen, I. N. Farstad, E. Domingo, D.N. Church, A. Nesbakken, N. A. Shepherd, I. Tomlinson, R. Kerr, M. Novelli, D. J.Kerr, and H. E. Danielsen, “Deep learning for prediction of colorectal cancer outcome:a discovery and validation study,”The Lancet, vol. 395, no. 10221, pp. 350–360, 2020.
[43] P. Courtiol, C. Maussion, M. Moarii, E. Pronier, S. Pilcer, M. Sefta, P. Manceron,S. Toldo, M. Zaslavskiy, N. L. Stang, N. Girard, O. Elemento, A. G. Nicholson, J.-Y.Blay, F. Galateau-Sall ́e, G. Wainrib, and T. Clozel, “Deep learning-based classificationof mesothelioma improves prediction of patient outcome,”Nature Medicine, vol. 25,no. 10, pp. 1519–1525, 2019.
[44] C. Saillard, B. Schmauch, O. Laifa, M. Moarii, S. Toldo, M. Zaslavskiy, E. Pronier,A. Laurent, G. Amaddeo, H. Regnault, D. Sommacale, M. Ziol, J.-M. Pawlotsky, S.Mul ́e, A. Luciani, G. Wainrib, T. Clozel, P. Courtiol, and J. Calderaro, “Predictingsurvival after hepatocellular carcinoma resection using deep-learning on histologicalslides,”Journal of Hepatology, vol. 73, 2020.41
[45] K.-H. Yu, C. Zhang, G. J. Berry, R. B. Altman, C. R ́e, D. L. Rubin, and M. Sny-der, “Predicting non-small cell lung cancer prognosis by fully automated microscopicpathology image features,”Nature Communications, vol. 7, no. 1, 2016.
[46] F. Isensee, P. F. Jaeger, S. A. A. Kohl, J. Petersen, and K. H. Maier-Hein, “nnU-Net:a self-configuring method for deep learning-based biomedical image segmentation,”Nature Methods, vol. 18, no. 2, pp. 203–211, 2020.
[47] D. Rueckert, L. Sonoda, C. Hayes, D. Hill, M. Leach, and D. Hawkes, “Nonrigidregistration using free-form deformations: application to breast MR images,”IEEETransactions on Medical Imaging, vol. 18, no. 8, pp. 712–721, 1999.
[48] C. Jost, J. Binek, J. C. Schuller, P. Bauerfeind, U. Metzger, B. Werth, J. Knuchel, J.-L.Frossard, P. Bertschinger, P. Brauchli, C. Meyenberger, and T. Ruhstaller, “Endosono-graphic radial tumor thickness after neoadjuvant chemoradiation therapy to predict re-sponse and survival in patients with locally advanced esophageal cancer: a prospectivemulticenter phase ll study by the Swiss Group for Clinical Cancer Research (SAKK75/02),”Gastrointestinal Endoscopy, vol. 71, no. 7, pp. 1114–1121, 2010.
[49] S. G. Swisher, M. Maish, J. J. Erasmus, A. M. Correa, J. A. Ajani, and R. Bresalier,“Utility of PET, CT, and EUS to identify pathologic responders in esophageal cancer,”Ann Thorac Surg, pp. 1152–1160, 2004.
[50] van Rossum, P., van Lier, A., van Vulpen, M., Reerink, O., Lagendijk, J., Lin, S., vanHillegersberg, R., Ruurda, J., Meijer, G. and Lips, I., Diffusion-weighted magnetic res-onance imaging for the prediction of pathologic response to neoadjuvant chemoradio-therapy in esophageal cancer.Radiotherapy and Oncology, vol. 115, no.2, pp. 163-170,2015.42
[51] Hamai, Y., Hihara, J., Emi, M., Furukawa, T., Murakami, Y., Nishibuchi, I., Nagata,Y., Ibuki, Y., Yamakita, I., Kurokawa, T. and Okada, M., Preoperative prediction of apathologic complete response of esophageal squamous cell carcinoma to neoadjuvantchemoradiotherapy.Surgery, vol. 164, no.1, pp.40-48, 2018.[52] Z. Yang, B. He, X. Zhuang, X. Gao, D. Wang, M. Li, Z. Lin, and R. Luo, “CT-based radiomic signatures for prediction of pathologic complete response in esophagealsquamous cell carcinoma after neoadjuvant chemoradiotherapy,”Journal of RadiationResearch, vol. 60, no. 4, pp. 538–545, 2019.
[53] S.-h. Chen, K. Ma, and Y.-f. Zheng, “Med3D: Transfer Learning for 3D MedicalImage Analysis,” arXiv preprint arXiv:1904.00625, 2019
[54] K. Hara, H. Kataoka, and Y. Satoh, “Can Spatiotemporal 3D CNNs Retrace the His-tory of 2D CNNs and ImageNet?,”2018 IEEE/CVF Conference on Computer Visionand Pattern Recognition, 2018.