光學同調斷層掃描(Optical Coherence Tomography, OCT)是一種光學成像方式。OCT可以分為兩類：頻域光學同調斷層掃描(Fourier Domain Optical Coherence Tomography, FD-OCT) 和時域光學同調斷層掃描(Time Domain Optical Coherence Tomography, TD-OCT)。由於FD-OCT在成像過程中，會產生共軛偽影、DC偽影和自相關偽影，這些偽影會造成成像失真，無法還原正確之光程差，因此偽影的消除是一個重要的議題。
近年來，深度學習(Deep learning, DL)被引用在多個領域，並取得相當程度上的成就，其中生成對抗網路(generative adversarial network, GAN)的討論度非常高。GAN網路是一種非監督式學習的系統，是透過兩個類神經網路相互對抗所組成的網路。本實驗室提出透過Pix2Pix GAN於單層結構樣品進行FD-OCT偽影上的去除。但當樣品變複雜時(單層/多層結構)，此時所得到的OCT A-scan影像必定會更加的複雜，將導致Pix2Pix GAN網路無法完美消除偽影。因此在本論文中，我們利用條件生成對抗網路，並設計新的骨幹類神經網路架構，以實現一個可用於多層全域式頻域光學同調斷層掃描系統的偽影消除器。

Optical coherence tomography (OCT) is an optical imaging method. OCT can be divided into two categories: Fourier-domain optical coherence tomography (FD-OCT)and time-domain optical coherence tomography (TD-OCT).Since FD-OCT will produce conjugate artifacts, DC artifacts, and autocorrelation artifacts during the imaging process, these artifacts will cause imaging distortion and cannot restore the correct optical path difference. Therefore, how to eliminate the artifacts is an important topic.
In recent years, deep learning (DL) has been applied in many fields and has achieved considerable achievements. Among them, the discussion of the generative adversarial network (GAN) is very high. GAN network is an unsupervised learning system. It is a network composed of two neural networks that compete against each other. Our laboratory proposed to use Pix2Pix GAN to remove FD-OCT artifacts on single-layer structure samples. But when the sample becomes complex (single-layer/multi-layer structure), the OCT A-scan image obtained at this time must be more complex, which will cause the Pix2Pix GAN network to be unable to perfectly eliminate artifacts. Therefore, in this thesis, we use conditional generative adversarial networks and design a new backbone neural network-like architecture to implement an artifact eliminator for multi-layer FD-OCT system.

中文摘要...i
英文摘要...ii
誌謝...iii
目錄...iv
表目錄...v
圖目錄...vi
第一章 緒論...1
1.1前言...1
1.2動機... 2
第二章 研究背景及相關知識...4
2.1 頻域光學同調斷層掃描...4
2.2 基於Pix2PixGAN的頻域光學同調斷層掃描...5
2.3 類神經網路介紹...6
2.3.1 深度神經網路...8
2.3.2 循環神經網路...8
2.3.3 卷積神經網路...9
2.3.4 生成對抗網路...14
2.3.5 pix2pix網路...15
2.3.6 U-net網路...16
第三章 研究內容與方法...17
3.1 CGAN網路之生成器...17
3.2 自定義的驗證標準...20
第四章 模擬結果及討論...25
4.1 CGAN網路上生成器為U-net網路的影像正確率...26
4.2 CGAN網路上生成器為U-ResNet網路的影像正確率...31
第五章 結論...36
5.1 論文結論...36
5.2 未來研究...36
參考文獻...37

[1]S. Asrani, L. Essaid, B. D. Alder, and C. Santiago-Turla, "Artifacts in Spectral-Domain Optical Coherence Tomography Measurements in Glaucoma," JAMA Ophthalmology, vol. 132, no. 4, pp. 396-402, 2014, doi: 10.1001/jamaophthalmol.2013.7974.
[2]J. Chhablani, T. Krishnan, V. Sethi, and I. Kozak, "Artifacts in optical coherence tomography," (in eng), Saudi J Ophthalmol, vol. 28, no. 2, pp. 81-7, Apr 2014, doi: 10.1016/j.sjopt.2014.02.010.
[3]J. Lin, S. Zhong, Q. Zhang, and W. Chen, "Five-Frame Variable Phase-Shifting Method for Full-Range Spectral-Domain Optical Coherence Tomography," Applied Sciences, vol. 8, no. 9, p. 1580, 2018. [Online]. Available: https://www.mdpi.com/2076-3417/8/9/1580.
[4]H.-C. Cheng, J.-F. Huang, and Y.-H. Hsieh, "Numerical analysis of one-shot full-range FD-OCT system based on orthogonally polarized light," Optics Communications, vol. 282, no. 14, pp. 3040-3045, 2009/07/15/ 2009, doi: https://doi.org/10.1016/j.optcom.2009.04.024.
[5]H.-C. Cheng and M.-S. Shiu, "Experimental demonstration of high-speed full-range Fourier domain optical coherence tomography imaging using orthogonally polarized light and a phase-shifting algorithm," Appl. Opt., vol. 51, no. 36, pp. 8762-8768, 2012/12/20 2012, doi: 10.1364/AO.51.008762.
[6]W. Bao, Y. Shen, T. Chen, P. Li, and Z. Ding, "High-speed high-precision and ultralong-range complex spectral domain dimensional metrology," Opt. Express, vol. 23, no. 9, pp. 11013-11022, 2015/05/04 2015, doi: 10.1364/OE.23.011013.
[7]Q. Zhang, S. Zhong, and J. Zhong, "Sine-modulated wavelength-independent full-range complex spectral optical coherence tomography with an ultra-broadband light source," Advances in Mechanical Engineering, vol. 7, no. 5, p. 1687814015588726, 2015, doi: 10.1177/1687814015588726.
[8]G. Liu et al., "Postprocessing algorithms to minimize fixed-pattern artifact and reduce trigger jitter in swept source optical coherence tomography," Opt. Express, vol. 23, no. 8, pp. 9824-9834, 2015/04/20 2015, doi: 10.1364/OE.23.009824.
[9]J. Lee, T. Yoon, and B. H. Lee, "Post-Processing Method for Image Reconstruction Enhancement in Integrating-Bucket-Based Full-Field Optical Coherence Tomography," Applied Sciences, vol. 10, no. 3, p. 830, 2020. [Online]. Available: https://www.mdpi.com/2076-3417/10/3/830.
[10]Y. Zhu and W. Gao, "Single-shot wavelength-independent phase-shifting method for full-field optical coherence tomography," Appl. Opt., vol. 58, no. 4, pp. 806-813, 2019/02/01 2019, doi: 10.1364/AO.58.000806.
[11]X. Liu, M. Ke, X. Yao, J. Chua, L. Schmetterer, and B. Tan, "Stable complex conjugate artifact removal in OCT using circularly polarized light as reference," Opt. Lett., vol. 45, no. 14, pp. 3977-3980, 2020/07/15 2020, doi: 10.1364/OL.395860.
[12]H.-C. Cheng and C.-T. Huang, "Measurement of Thickness and Refractive Index of Optical Samples Simultaneously Using Full-Range One-Shot Frequency-Domain Optical Coherence Tomography," Fiber and Integrated Optics, vol. 34, no. 3, pp. 145-156, 2015/05/04 2015, doi: 10.1080/01468030.2015.1044678.
[13]S. K. Devalla et al., "A Deep Learning Approach to Denoise Optical Coherence Tomography Images of the Optic Nerve Head," Scientific Reports, vol. 9, no. 1, p. 14454, 2019/10/08 2019, doi: 10.1038/s41598-019-51062-7.
[14]B. Buchroithner, A. Prylepa, P. J. Wagner, S. E. Schausberger, D. Stifter, and B. Heise, "Full-field optical coherence tomography in a balanced detection mode," Appl. Opt., vol. 57, no. 29, pp. 8705-8710, 2018/10/10 2018, doi: 10.1364/AO.57.008705.
[15]H. Cheong et al., "DeshadowGAN: A Deep Learning Approach to Remove Shadows from Optical Coherence Tomography Images," Translational Vision Science & Technology, vol. 9, no. 2, pp. 23-23, 2020, doi: 10.1167/tvst.9.2.23.
[16]C. H. Lin et al., "Denoising Performance Evaluation of Automated Age-Related Macular Degeneration Detection on Optical Coherence Tomography Images," IEEE Sensors Journal, vol. 21, no. 1, pp. 790-801, 2021, doi: 10.1109/JSEN.2020.3014254.
[17]H. Abdelmotaal, A. A. Abdou, A. F. Omar, D. M. El-Sebaity, and K. Abdelazeem, "Pix2pix Conditional Generative Adversarial Networks for Scheimpflug Camera Color-Coded Corneal Tomography Image Generation," Translational Vision Science & Technology, vol. 10, no. 7, pp. 21-21, 2021, doi: 10.1167/tvst.10.7.21 %J Translational Vision Science & Technology.
[18]I. Goodfellow et al., "Generative adversarial networks," Communications of the ACM, vol. 63, no. 11, pp. 139-144, 2020.
[19]M. Mirza and S. Osindero, "Conditional Generative Adversarial Nets," ArXiv, vol. abs/1411.1784, 2014.
[20]H. Cheong et al., "OCT-GAN: single step shadow and noise removal from optical coherence tomography images of the human optic nerve head," Biomed. Opt. Express, vol. 12, no. 3, pp. 1482-1498, 2021/03/01 2021, doi: 10.1364/BOE.412156.
[21]Q. Hao et al., "High signal-to-noise ratio reconstruction of low bit-depth optical coherence tomography using deep learning," Journal of Biomedical Optics, vol. 25, no. 12, p. 123702, 2020. [Online]. Available: https://doi.org/10.1117/1.JBO.25.12.123702.
[22]A. Guo, L. Fang, M. Qi, and S. Li, "Unsupervised Denoising of Optical Coherence Tomography Images With Nonlocal-Generative Adversarial Network," IEEE Transactions on Instrumentation and Measurement, vol. 70, pp. 1-12, 2021, doi: 10.1109/TIM.2020.3017036.
[23]Y. Huang et al., "Noise-Powered Disentangled Representation for Unsupervised Speckle Reduction of Optical Coherence Tomography Images," IEEE Transactions on Medical Imaging, vol. 40, no. 10, pp. 2600-2614, 2021, doi: 10.1109/TMI.2020.3045207.
[24]T. Zhang et al., "Noise Adaptation Generative Adversarial Network for Medical Image Analysis," IEEE Transactions on Medical Imaging, vol. 39, no. 4, pp. 1149-1159, 2020, doi: 10.1109/TMI.2019.2944488.
[25] P. Isola, J. Y. Zhu, T. Zhou, and A. A. Efros, "Image-to-Image Translation with Conditional Adversarial Networks," in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 21-26 July 2017 2017, pp. 5967-5976, doi: 10.1109/CVPR.2017.632.
[26]C. M. Huang, E. Wijanto, and H. C. Cheng, "Applying a Pix2Pix Generative Adversarial Network to a Fourier-Domain Optical Coherence Tomography System for Artifact Elimination," IEEE Access, vol. 9, pp. 103311-103324, 2021, doi: 10.1109/ACCESS.2021.3098865.
[27]C. Zheng et al., "Assessment of Generative Adversarial Networks for Synthetic Anterior Segment Optical Coherence Tomography Images in Closed-Angle Detection," Translational Vision Science & Technology, vol. 10, no. 4, pp. 34-34, 2021, doi: 10.1167/tvst.10.4.34.
[28] F. Ning, "Multi-style Migration of Chinese Characters based on Self-attention Mechanism and StarGAN v2," in 2022 3rd International Conference on Computer Vision, Image and Deep Learning & International Conference on Computer Engineering and Applications (CVIDL & ICCEA), 20-22 May 2022 2022, pp. 357-361, doi: 10.1109/CVIDLICCEA56201.2022.9825460.
[29]J. A. Izatt and M. A. Choma, "Theory of Optical Coherence Tomography," in Optical Coherence Tomography: Technology and Applications, W. Drexler and J. G. Fujimoto Eds. Berlin, Heidelberg: Springer Berlin Heidelberg, 2008, pp. 47-72.
[30] Y. Kochura, S. Stirenko, and Y. Gordienko, "Comparative performance analysis of neural networks architectures on H2O platform for various activation functions," in 2017 IEEE International Young Scientists Forum on Applied Physics and Engineering (YSF), 2017: IEEE, pp. 70-73.
[31]G. Melki, "Fast Online Training of L1 Support Vector Machines," 2016.
[32]R. Devi D and d. I, "Deep Learning Methods for Big Data Analytics," 10/01 2021.
[33]J. Feng, X. He, Q. Teng, C. Ren, H. Chen, and Y. Li, "Reconstruction of porous media from extremely limited information using conditional generative adversarial networks," Physical Review E, vol. 100, 09/16 2019, doi: 10.1103/PhysRevE.100.033308.
[34]G. Roffo, "Ranking to Learn and Learning to Rank: On the Role of Ranking in Pattern Recognition Applications," 06/01 2017.
[35]A. Mino and G. Spanakis, LoGAN: Generating Logos with a Generative Adversarial Neural Network Conditioned on color. 2018.
[36] O. Ronneberger, P. Fischer, and T. Brox, "U-net: Convolutional networks for biomedical image segmentation," in International Conference on Medical image computing and computer-assisted intervention, 2015: Springer, pp. 234-241.
[37]K. He, X. Zhang, S. Ren, and J. Sun, "Deep Residual Learning for Image Recognition," p. arXiv:1512.03385. [Online]. Available: https://ui.adsabs.harvard.edu/abs/2015arXiv151203385H