跳到主要內容

臺灣博碩士論文加值系統

(44.192.48.196) 您好!臺灣時間:2024/06/14 17:16
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

: 
twitterline
研究生:楊竣皓
研究生(外文):Jun-Hao Yang
論文名稱:探討自監督學習方法於Cryo-EM影像去噪的潛力
論文名稱(外文):Unveiling the Potential of Self-Supervised Learning for Cryo-EM Image Denoising
指導教授:鍾思齊
指導教授(外文):Chung, Szu - Chi
學位類別:碩士
校院名稱:國立中山大學
系所名稱:應用數學系研究所
學門:數學及統計學門
學類:數學學類
論文種類:學術論文
論文出版年:2023
畢業學年度:111
語文別:中文
論文頁數:58
中文關鍵詞:自監督學習Cryo-EM影像影像去雜訊U-Net超參數最佳化
外文關鍵詞:Self-supervised LearningCryo-EM imagesImage denoisingU-NetHyperparameter optimization
相關次數:
  • 被引用被引用:0
  • 點閱點閱:52
  • 評分評分:
  • 下載下載:10
  • 收藏至我的研究室書目清單書目收藏:0
本研究旨在探討利用深度學習自監督的方式應用於Cryo-EM 蛋白質影像進行去雜訊。冷凍電子顯微術( Cryo-electron microscopy, Cryo-EM)是結構生物學研究最重要的新技術之一,近年來Cryo-EM更成功揭發蛋白質在原子尺度下的結構。但由於儀器和蛋白質本身的限制,所以圖像中的雜訊是很高的,這對於要重建回三維結構的蛋白質會是很大的挑戰,因此影像的去雜訊擔任了一個很重要的任務。相比於傳統的小波轉換、總變差及BM3D的去雜訊方法及早期的神經網路方法需要乾淨的影像或是吵雜的圖像對進行訓練的限制,我們使用四個文獻中新提出的自監督學習架構:Noise2Void, Self2Self, Noise2Same, Neighbor2Neighbor這些訓練時只需要吵雜的Cryo-EM影像的方法,並探討這些方法的調整及超參數對於去雜訊效果的影響。研究結果顯示我們的方法不需要知道雜訊水平也不需要乾淨影像與成對吵雜影像便可以成功完成去雜訊,且我們的結果可以與傳統方法相近甚至是超越。最後希望未來可以使用我們的方法在真實資料集並應用在不同Cryo-EM任務上。
The purpose of this study is to investigate the application of self-supervised methods for denoising protein images. Cryo-electron microscopy (Cryo-EM) is one of the most important new techniques in structural biology research, and in recent years, Cryo-EM has successfully revealed protein structures at the atomic scale. However, due to limitations in instruments and the proteins themselves, the images obtained often contain high levels of noise, posing a significant challenge for reconstructing the three-dimensional structure of proteins. Therefore, image denoising plays a crucial role in this process. In comparison to traditional denoising methods such as wavelet, total variation, BM3D, and early neural network approaches that require either clean images or pairs of noisy images for training, we utilize four newly proposed self-supervised learning frameworks from the literature: Noise2Void, Self2Self, Noise2Same, and Neighbor2Neighbor. These methods only require noisy Cryo-EM images during training, and we investigate the effects of adjusting these methods and their hyperparameters on denoising performance. The results of the study demonstrate that our approach can successfully denoise Cryo-EM images without the need for knowing the noise level or clean images or paired noisy images. Moreover, our results are comparable to or even surpass traditional methods. Finally, we hope to apply our method to real datasets and various Cryo-EM tasks in the future.
論文審定書 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i 摘要...........................................................................ii Abstract.......................................................................iii Chapter1 Introduction.......................................................1
1.1 介紹Cryo-EM...........................................................1 1.2 去雜訊對Cryo-EM的重要性.............................................1 1.3 研究動機與目標..........................................................2
Chapter2 Previouswork.....................................................3
2.1 WaveletDenoising........................................................3
2.2 Total Variation Denoising.................................................4
2.3 BM3D...................................................................4
2.4 DenoisingCNN..........................................................5
2.5 Noise2Noise..............................................................6
2.6 Noise2Self...............................................................7
Chapter3 Method............................................................9 3.1 Noise2Void...............................................................9 3.2 Self2Self................................................................10 3.3 Noise2Same.............................................................12 3.4 Neighbor2Neighbor.....................................................13
Chapter4 Experiments......................................................17 4.1 Synthetic data generation................................................17 4.2 Evaluationmetric.......................................................18
4.2.1 PSNR................................................................18
4.2.2 SSIM................................................................18 4.3 超參數討論.............................................................19 4.3.1 N2V.................................................................19 4.3.2 S2S-one..............................................................22 4.3.3 S2S-sam32...........................................................24 4.3.4 S2S-bs32.............................................................26 4.3.5 N2Same..............................................................28 4.3.6 Nei2Nei..............................................................31 4.4 在70Sribosome各種去雜訊方法結果和討論.............................34 Chapter 5 Conclusion and Future Work......................................40 5.1 結論....................................................................40 5.2 未來展望...............................................................40 References....................................................................42 Chapter6 Appendix.........................................................45
6.1 附錄一:N2V...........................................................45
6.2 附錄二:S2S-one ....................................................... 45
6.3 附錄三:S2S-sam32.....................................................46
6.4 附錄四:S2S-bs32 ...................................................... 46
6.5 附錄五:N2Same.......................................................47
6.6 附錄六:Nei2Nei........................................................47
6.7 附錄七:真實影像去雜訊 ............................................... 48
[1] A. Singer and F. J. Sigworth, “Computational methods for single-particle elec- tron cryomicroscopy,” Annual review of biomedical data science, vol. 3, pp. 163–190, 2020.
[2] J. Ouyang, Z. Liang, C. Chen, Z. Fu, Y. Zhang, and H. Liu, “Cryo-electron microscope image denoising based on the geodesic distance,” BMC Structural Biology, vol. 18, pp. 1–17, 2018.
[3] T. Bhamre, T. Zhang, and A. Singer, “Denoising and covariance estimation of single particle cryo-em images,” Journal of structural biology, vol. 195, no. 1, pp. 72–81, 2016.
[4] H. Lei and Y. Yang, “Cdae: a cascade of denoising autoencoders for noise reduc- tion in the clustering of single-particle cryo-em images,” Frontiers in genetics, vol. 11, p. 627746, 2021.
[5] H. Li, H. Zhang, X. Wan, Z. Yang, C. Li, J. Li, R. Han, P. Zhu, and F. Zhang, “Noise-transfer2clean: denoising cryo-em images based on noise modeling and transfer,” Bioinformatics, vol. 38, no. 7, pp. 2022–2029, 2022.
[6] T. Bepler, K. Kelley, A. J. Noble, and B. Berger, “Topaz-denoise: general deep denoising models for cryoem and cryoet,” Nature communications, vol. 11, no. 1, p. 5208, 2020.
[7] L. Fan, F. Zhang, H. Fan, and C. Zhang, “Brief review of image denoising techniques,” Visual Computing for Industry, Biomedicine, and Art, vol. 2, no. 1, pp. 1–12, 2019.
[8] B. Goyal, A. Dogra, S. Agrawal, B. S. Sohi, and A. Sharma, “Image denois- ing review: From classical to state-of-the-art approaches,” Information fusion, vol. 55, pp. 220–244, 2020.
[9] J. Lehtinen, J. Munkberg, J. Hasselgren, S. Laine, T. Karras, M. Aittala, and T. Aila, “Noise2noise: Learning image restoration without clean data,” arXiv preprint arXiv:1803.04189, 2018.
[10] J. Batson and L. Royer, “Noise2self: Blind denoising by self-supervision,” in International Conference on Machine Learning. PMLR, 2019, pp. 524–533.
[11] A. Krull, T.-O. Buchholz, and F. Jug, “Noise2void-learning denoising from single noisy images,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 2129–2137.
[12] O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” in Medical Image Computing and Computer- Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III 18. Springer, 2015, pp. 234–241.
[13] L. N. Smith, “A disciplined approach to neural network hyper-parameters: Part 1–learning rate, batch size, momentum, and weight decay,” arXiv preprint arXiv:1803.09820, 2018.
[14] Y. Quan, M. Chen, T. Pang, and H. Ji, “Self2self with dropout: Learning self-supervised denoising from single image,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 1890–1898.
[15] G. Liu, F. A. Reda, K. J. Shih, T.-C. Wang, A. Tao, and B. Catanzaro, “Image inpainting for irregular holes using partial convolutions,” in Proceedings of the European conference on computer vision (ECCV), 2018, pp. 85–100.
[16] Y. Xie, Z. Wang, and S. Ji, “Noise2same: Optimizing a self-supervised bound for image denoising,” Advances in Neural Information Processing Systems, vol. 33, pp. 20 320–20 330, 2020.
[17] T. Huang, S. Li, X. Jia, H. Lu, and J. Liu, “Neighbor2neighbor: Self-supervised denoising from single noisy images,” pp. 14 781–14 790, 2021.
[18] S. H. Scheres, “Semi-automated selection of cryo-em particles in relion-1.3,” Journal of structural biology, vol. 189, no. 2, pp. 114–122, 2015.
[19] A. Punjani, J. L. Rubinstein, D. J. Fleet, and M. A. Brubaker, “cryosparc: algorithms for rapid unsupervised cryo-em structure determination,” Nature methods, vol. 14, no. 3, pp. 290–296, 2017.
[20] E. F. Pettersen, T. D. Goddard, C. C. Huang, G. S. Couch, D. M. Greenblatt, E. C. Meng, and T. E. Ferrin, “Ucsf chimera—a visualization system for ex- ploratory research and analysis,” Journal of computational chemistry, vol. 25, no. 13, pp. 1605–1612, 2004.
[21] S.-C. Chung, H.-H. Lin, P.-Y. Niu, S.-H. Huang, I.-P. Tu, and W.-H. Chang, “Pre-pro is a fast pre-processor for single-particle cryo-em by enhancing 2d classification,” Communications biology, vol. 3, no. 1, p. 508, 2020.
[22] E. Palovcak, D. Asarnow, M. G. Campbell, Z. Yu, and Y. Cheng, “Enhancing the signal-to-noise ratio and generating contrast for cryo-em images with convolutional neural networks,” IUCrJ, vol. 7, no. 6, pp. 1142–1150, 2020.
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top