計算照明高通量生物醫學成像顯微鏡系統

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：57

、訪客IP：18.217.246.148

姓名

郭俊甫(Chun-Fu Kuo) 查詢紙本館藏

畢業系所

生醫科學與工程學系

論文名稱

計算照明高通量生物醫學成像顯微鏡系統
(Computational illumination-based microscope system for high-throughput biomedical imaging)

相關論文

★ 具生物沾粘性免疫奈米磁珠之電化學平台於急性冠心病標誌物檢測	★ 離子液體應用於脂溶性蛋白之快速萃取及檢測
★ 深度學習計算成像系統於生物醫學顯微影像之重建與分析	★ 磁電化學免疫分析系統於新型冠狀病毒感染檢測之研製
★ 表面增強拉曼散射探針及微流道系統用於癌症細胞及外泌體表面生物標記物的多重檢測	★ 透過深度學習神經網路進行無透鏡數位全像顯微系統之影像重建與相位恢復
★ 通過多醣表面修飾增強隱形眼鏡的親水性和抗菌性能	★ 菲涅耳數位全像顯微系統於全血細胞分析之研製
★ 遮罩區域卷積類神經網路於醫學影像物件偵測分析應用	★ 基於深度學習之高通量顯微影像系統於血液分析
★ 可功能化編碼之表面增強拉曼光譜標籤探針於高靈敏與多重生醫分子檢測	★ 高通量計算顯微影像系統之研製於生物醫學成像與分析
★ 功能性抗生物沾黏單層膜於冠狀動脈心血管疾病標誌物之檢測應用	★ 鼻咽癌外泌小體藉由EB病毒潛伏膜蛋白1促進巨噬細胞免疫抑制型分化

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

至系統瀏覽論文 ( 永不開放)

摘要(中)

傅立葉疊層圖像與人工智慧的結合為解析度改善提供了一個強大的工具。傅立葉疊
層圖像 (Fourier Ptychography,FP)利用傅立葉變換原理，藉由不同激發光陣列，以提供不
同角度之點光源，能夠增強圖像的細節和結構。同時，人工智慧 Artificial Intelligence (AI)技術如深度學習和卷積神經網絡 (Convolution Neural Network,CNN)已在圖像處理中取得
巨大成功。
通過結合這兩種技術，我們能夠更進一步改善圖像的解析度，相比於傳統顯微影像
系統，我們擁有與高倍率相同的解析度與低倍率相同視場範圍 (Field Of View,FOV)。具
體而言，我們可以使用傅立葉疊層圖像提取出高頻細節和低頻結構，然後利用人工智慧
算法來重建缺失的細節。這種結合可以通過深度學習模型對圖像進行超分辨率處理，以
提高圖像的解析度和品質。
人工智慧算法能夠學習從低解析度圖像到高解析度圖像之間的映射關係，從而在重
建過程中增加細節並恢復圖像的清晰度。這種結合也為圖像重建提供了更多的彈性，因
為人工智慧算法可以根據具體應用和需求進行訓練和調整。
總之，傅立葉疊層圖像搭配人工智慧技術的結合為解析度改善帶來了巨大的潛力。
這種結合能夠提高圖像的解析度，恢復細節並提供更好的視覺品質。隨著人工智慧技術
的不斷發展和傅立葉疊層圖像的應用擴大，這一結合將在許多領域，如醫學影像、監控
系統和衛星圖像等方面帶來更多的突破和創新。

摘要(英)

The combination of Fourier Ptychography (FP) and Artificial Intelligence (AI) provides a powerful tool for improving resolution. Fourier Ptychography utilizes the principles of Fourier transformation and captures images under different illuminations using an array of point sources at different angles, which enhances the details and structures in the image. On the other hand, AI techniques such as deep learning and Convolutional Neural Networks (CNN) have achieved significant success in image processing.
By combining these two techniques, we can further improve the resolution of the image. Compared to traditional microscopy systems, Fourier Ptychography with AI enables us to achieve the same resolution as high magnification, while maintaining the same field of view. Specifically, Fourier Ptychography is used to extract high-frequency details and low-frequency structures, and AI algorithms are employed to reconstruct missing details. This combination can be achieved through deep learning models for super-resolution processing, thereby enhancing the resolution and quality of the image.
AI algorithms can learn the mapping relationship between low-resolution and high-resolution images, thereby increasing the details and restoring the clarity of the image during the reconstruction process. This combination also provides more flexibility in image reconstruction, as AI algorithms can be trained and adjusted according to specific applications and requirements.
In conclusion, the combination of Fourier Ptychography and AI offers great potential for resolution improvement. This combination enhances the resolution, restores details, and provides better visual quality in the image. With the continuous development of AI technology and the expanding applications of Fourier Ptychography, this integration will bring about further breakthroughs and innovations in various fields such as medical imaging, surveillance systems, and satellite imagery.

關鍵字(中)

★ 傅立葉疊層圖像
★ 計算成像系統
★ 深度學習
★ 人工智慧

關鍵字(英)

★ Fourier Ptychography
★ Computational Imaging System
★ Deep Learning
★ Artificial Intelligence

論文目次

摘要 ......................................................................................................................................... i
Abstract.................................................................................................................................. ii
致謝 ....................................................................................................................................... iii
目錄 ........................................................................................................................................ 1
第一章緒論 ......................................................................................................................... 5
1.1 研究方法與動機 ................................................................................................. 5
1.2 論文架構 ............................................................................................................. 6
第二章文獻探討 ................................................................................................................. 7 2.1 傅立葉疊層成像 ................................................................................................. 7
2.1-1 傅立葉轉換 ................................................................................................... 7
2.1-2 成像系統原理和限制 ................................................................................. 10
2.1-3 傅立葉疊層成像原理 ................................................................................. 12
2.1-4 傅立葉疊層成像技術 ................................................................................. 13
2.2 人工智慧 ........................................................................................................... 18
2.2-1 機器學習 ..................................................................................................... 18
2.2-2 深度學習 ..................................................................................................... 19
2.2-3 卷積神經網絡 ............................................................................................. 20
2.2-4 超解析度生成對抗網絡 ............................................................................. 21
第三章實驗方法 ............................................................................................................. 25
3.1 研究方法與架構................................................................................................ 25
3.2 挑選硬體及架設傅立葉疊層成像系統與重建 ................................................ 25
3.2-1 系統硬體參數 ............................................................................................. 25
3.2-2 架設系統硬體設備 ..................................................................................... 26
3.2-3 低解析度圖像擷取選擇與流程 ................................................................. 28
3.2-4 重建程式迭代順序 ..................................................................................... 30
3.3 影像提升之深度學習模型 ................................................................................ 30
3.3-1 超解析度生成對抗網絡模型(SRGAN) ..................................................... 31
3.3-2 超解析度生成對抗網絡模型的調整.......................................................... 31
2
第四章實驗結果與討論.................................................................................................. 33
4-1 傅立葉疊層成像實驗結果 ................................................................................ 33
4-1-1 資料收集 ..................................................................................................... 33
4-1-2 結果比較 ..................................................................................................... 40
4-2 深度學習影像增強 ........................................................................................... 46
4-2-1 深度學習資料集 ......................................................................................... 46
4-2-2 深度學習結果與討論 ................................................................................. 48
第五章結論 ..................................................................................................................... 52
5-1 結論 ................................................................................................................... 52
5-2 未來展望 ............................................................................................................ 52
參考文獻Reference ............................................................................................................. 53

參考文獻

1. G. Zheng, R. Horstmeyer and C. Yang, "Wide-field high-resolution Fourier ptychographic microscopy", Nature Photon., vol. 7, pp. 739-745, 2013.
2. Lohmann,A.W.,et al.,Space-bandwith product of optical signals and systems.JOSAA, 1996.13(3):p.470-473
3. Christian Ledig,,et al., Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network arXiv:1609.04802,2017
4. A. M. Maiden and J. M. Rodenburg, “An improved ptychographical phase retrieval algorithm for diffractive imaging,”Ultramicroscopy 109(10), 1256–1262 (2009).
5. Fienup J R 1982 Phase retrieval algorithms: a comparison Appl. Opt. 21 2758–69
6. Elser V 2003 Phase retrieval by iterated projections JOSA A 20 40–55
7. Maiden A M and Rodenburg J M 2009 An improved ptychographical phase retrieval algorithm for diffractive imaging Ultramicroscopy 109 1256–62
8. Fienup J R 1987 Reconstruction of a complex-valued object from the modulus of its Fourier transform using a support constraint JOSA A 4 118–23
9. K. S. Youngworth and T. G Brown, Focusing of high numerical aperture cylindrical-vector beams, Optics Express, Vol. 7, Issue 2, pp. 77-87 (2000)
10. Rimas Juškaitis, Characterizing High Numerical Aperture Microscope Objective Lenses, Part of the Springer Series in Optical Sciences book series (SSOS,volume 87)
11. Xiaoze Ou, Roarke Horstmeyer, Guoan Zheng, and Changhuei Yang, High numerical aperture Fourier ptychography: principle, implementation and characterization, Optics Express Vol. 23, Issue 3, pp. 3472-3491 (2015)
12. Ou X, Horstmeyer R, Yang C and Zheng G 2013 Quantitative phase imaging via Fourier ptychographic microscopy Opt. Lett. 38 4845–8
13. Bian L, Suo J, Situ G, Zheng G, Chen F and Dai Q 2014 Content adaptive illumination for Fourier ptychography Opt. Lett. 39 6648–51
14. Dong S, Bian Z, Shiradkar R and Zheng G 2014 Sparsely sampled Fourier ptychography Opt. Exp. 22 5455–64
54
15. Dong S, Shiradkar R, Nanda P and Zheng G 2014 Spectral multiplexing and coherent-state decomposition in Fourier ptychographic imaging Biomedical Opt. Exp. 5 1757–67
16. Jiasong Sun, Qian Chen, Yuzhen Zhang, and Chao Zuo, Sampling criteria for Fourier ptychographic microscopy in object space and frequency space, Optics Express Vol. 24, Issue 14, pp. 15765-15781 (2016)
17. Dick, S. (2019) Artificial Intelligence. Harvard Data Science Review, 1, 1-9.
18. Jordan, M. I., & Mitchell, T. M. (2015). Machine learning: Trends, perspectives, and prospects. Science, 349(6245), 255–260
19. Keiron O′Shea, Ryan Nash, An Introduction to Convolutional Neural Networks, arXiv:1511.08458
20. Ian J. Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, Yoshua Bengio, Generative Adversarial Networks, arXiv:1406.2661
21. Wenlong Zhang, Yihao Liu, Chao Dong, Yu Qiao, RankSRGAN: Generative Adversarial Networks with Ranker for Image Super-Resolution,arXiv:1908.06382
22. Jun-Hong Huang, SIR-SRGAN: Super-Resolution Generative Adversarial Network with Self-Interpolation Ranker
23. Timothy O. Hodson, Mean Squared Error, Deconstructed, Gupta, H. V., Kling, H., Yilmaz, K. K., & Martinez, G. F. (2009). Decomposition of the mean squared error and NSE performance criteria: Implications for improving hydrological modelling. Journal of Hydrology, 377(1–2), 80– 91
24. Jim Nilsson, Understanding SSIM, arXiv:2006.13846

指導教授

黃貞翰(Chen-Han Huang)

審核日期

2023-8-21

推文