光片荧光显微图像的亮度校正与条纹伪影去除

光片荧光显微镜是一种新型的激光扫描式的荧光显微镜，它可以对生物样本进行高分辨率的三维成像，被广泛应用于各种生物学和医学研究中，如大脑的三维成像。本文关注的是光片荧光显微镜普遍存在的照度失真现象，该现象带来了 图像整体亮度不均匀和存在条纹伪影两类问题。图像亮度校正方法分为两类，前瞻性方法和回顾性方法。前瞻性方法需要更改显微镜拍摄流程，不具有通用性；回顾性方法无法校正形状不规则或者只占一部分区域的样本的图像亮度。条纹伪影去除方案可分为光学方案和算法方案两种，光学方案需要更改显微镜结构，也不具有通用性；而现有的算法方案图像处理时间长，效果较差，不适合高通量显微镜的需求。

针对光片荧光显微图像亮度校正，本文设计了一种自适应图像亮度校正算法， 它可以自动识别图像中样本的区域，并提取相关像素来计算亮度校正参数。和其它亮度校正算法相比，该算法可以校正图像中任意形状和大小样本的亮度，不需要样本占满图像的整个区域。实验证明，该算法可以有效修复投影图和三维重构图的拼接条纹，均衡荧光细胞亮度并提高细胞分割的准确率。

针对光片荧光显微图像条纹伪影去除，本文在没有无条纹真实图像（即形式 上无监督的情况下制作了相应的数据集，并训练出了一种卷积神经网络模型。实验证明，该模型可以有效去除原始图的条纹伪影，比其它算法的处理效果更好，处理时间更短。为了增强生成网络的泛化能力，本文提出了一种生成对抗网络模型。 该模型可以在数据集中标签图像的条纹伪影没有去除干净的情况下，使训练的生 成网络的条纹伪影去除效果比标签图像更好。去除条纹伪影解决了图像退化问题， 恢复图像中样本的真实结构。

Light Sheet Fluorescence Microscopy is a new type of fluorescence microscope that can achieve high-resolution 3D imaging of biological samples. It has been widely used in various biological and medical research fields, such as 3D brain imaging. This article focuses on the illumination distortion phenomenon commonly found in these microscopes, which results in two types of problems: uneven image illumination and the presence of striped artifacts. There are two types of methods for correcting the image brightness: prospective methods and retrospective methods. Prospective methods require changes to the microscope's imaging process and are not universally applicable, while retrospective methods are unable to correct the illumination of samples with irregular shapes or those occupying only a portion of the field of view. Stripe artifact removal solutions can be divided into two categories: optical and algorithmic. Optical solutions require changes to the microscope's hardware and are not universally applicable, whereas existing algorithmic solutions have long processing times and poor results, making them unsuitable for high-throughput microscopy applications.

To address image illumination correction, this article proposes an adaptive image illumination correction algorithm that can automatically identify the sample region in the image and extract relevant pixels to calculate the illumination correction parameters. Compared to other illumination correction algorithms, this algorithm can correct the illumination of any size or shape sample in the image without requiring the sample to occupy the entire field of view. Experimental results show that this algorithm can effectively repair projection images and 3D reconstruction images with stripe artifacts, equalize cell fluorescence illumination, and improve cell segmentation accuracy.

For stripe artifact removal of image, this article created a corresponding dataset in the absence of unstriped ground truth images (i.e., in a form of unsupervised learning) and trained a convolutional neural network. Experimental results show that this model can effectively remove stripe artifacts from original images with better results and shorter processing times than other algorithms. To enhance the generative network's generalization ability, this article proposes a generative adversarial network that can improve the stripe artifact removal performance of the trained generative network compared to the ground truth images in the dataset where the label image stripe artifacts are not completely removed. Removing stripe artifacts solves the problem of image degradation and restores the true structure of the samples in the image.

[1] POWER R M, HUISKEN J. A guide to light-sheet fluorescence microscopy for multiscale imaging[J]. Nature methods, 2017, 14(4): 360-373.
[2] CHAKRABORTY T, DRISCOLL M K, JEFFERY E, et al. Light-sheet microscopy of cleared tissues with isotropic, subcellular resolution[J]. Nature Methods, 2019, 16(11): 1109-1113.
[3] CHEN B C, LEGANT W R, WANG K, et al. Lattice light-sheet microscopy: imaging molecules to embryos at high spatiotemporal resolution[J]. Science, 2014, 346(6208): 1257998.
[4] LAGACHE T, GRASSART A, DALLONGEVILLE S, et al. Mapping molecular assemblies with fluorescence microscopy and object-based spatial statistics[J]. Nature communications, 2018, 9(1): 698.
[5] PAMPALONI F, BERGE U, MARMARAS A, et al. Tissue-culture light sheet fluorescence mi- croscopy (TC-LSFM) allows long-term imaging of three-dimensional cell cultures under con- trolled conditions[J]. Integrative Biology, 2014, 6(10): 988-998.
[6] FIOLKA R. Light-sheet microscopy at high resolution[J]. Nature biotechnology, 2021, 39(11): 1345-1346.
[7] KELLER P J, AHRENS M B. Visualizing whole-brain activity and development at the single- cell level using light-sheet microscopy[J]. Neuron, 2018, 85(3): 462-483.
[8] DERYCKERE A, STYFHALS R, VIDAL E A, et al. A practical staging atlas to study embry- onic development of Octopus vulgaris under controlled laboratory conditions[J]. BMC Devel- opmental Biology, 2020, 20: 1-18.
[9] YALCIN H C, AMINDARI A, BUTCHER J T, et al. Heart function and hemodynamic analysis for zebrafish embryos[J]. Developmental Dynamics, 2017, 246(11): 868-880.
[10] MIR M, REIMER A, STADLER M, et al. Single molecule imaging in live embryos using lattice light-sheet microscopy[J]. Nanoscale Imaging: Methods and Protocols, 2018: 541-559.
[11] MEMEO R, PAIÈ P, SALA F, et al. Automatic imaging of Drosophila embryos with light sheet fluorescence microscopy on chip[J]. Journal of Biophotonics, 2021, 14(3): e202000396.
[12] BLUTKE A, SUN N, XU Z, et al. Light sheet fluorescence microscopy guided MALDI-imaging mass spectrometry of cleared tissue samples[J]. Scientific Reports, 2020, 10(1): 14461.
[13] AMICH J, MOKHTARI Z, STROBEL M, et al. Three-dimensional light sheet fluorescence microscopy of lungs to dissect local host immune-Aspergillus fumigatus interactions[J]. MBio, 2020, 11(1): e02752-19.
[14] KLINGBERG A, HASENBERG A, LUDWIG-PORTUGALL I, et al. Fully automated eval- uation of total glomerular number and capillary tuft size in nephritic kidneys using lightsheet microscopy[J]. Journal of the American Society of Nephrology, 2017, 28(2): 452-459.
[15] FANG C, YU T, CHU T, et al. Minutes-timescale 3D isotropic imaging of entire organs at subcellular resolution by content-aware compressed-sensing light-sheet microscopy[J]. Nature Communications, 2021, 12(1): 107.
[16] CAI R, PAN C, GHASEMIGHARAGOZ A, et al. Panoptic imaging of transparent mice reveals whole-body neuronal projections and skull–meninges connections[J]. Nature neuroscience, 2019, 22(2): 317-327.
[17] WANG W, ZHANG Y, HUI H, et al. The effect of endothelial progenitor cell transplantation on neointimal hyperplasia and reendothelialisation after balloon catheter injury in rat carotid arteries[J]. Stem Cell Research & Therapy, 2021, 12(1): 1-12.
[18] LUGO-HERNANDEZ E, SQUIRE A, HAGEMANN N, et al. 3D visualization and quantifica- tion of microvessels in the whole ischemic mouse brain using solvent-based clearing and light sheet microscopy[J]. Journal of Cerebral Blood Flow & Metabolism, 2017, 37(10): 3355-3367.
[19] FRANÇA C M, RIGGERS R, MUSCHLER J L, et al. 3D-imaging of whole neuronal and vas- cular networks of the human dental pulp via CLARITY and light sheet microscopy[J]. Scientific reports, 2019, 9(1): 10860.
[20] LAZZARI G, VINCIGUERRA D, BALASSO A, et al. Light sheet fluorescence microscopy versus confocal microscopy: In quest of a suitable tool to assess drug and nanomedicine pene- tration into multicellular tumor spheroids[J]. European Journal of Pharmaceutics and Biophar- maceutics, 2019, 142: 195-203.
[21] NEUFERT C, HEICHLER C, BRABLETZ T, et al. Inducible mouse models of colon cancer for the analysis of sporadic and inflammation-driven tumor progression and lymph node metastasis [J]. Nature Protocols, 2021, 16(1): 61-85.
[22] TANAKA N, KACZYNSKA D, KANATANI S, et al. Mapping of the three-dimensional lym- phatic microvasculature in bladder tumours using light-sheet microscopy[J]. British Journal of Cancer, 2018, 118(7): 995-999.
[23] ALLADIN A, CHAIBLE L, GARCIA DEL VALLE L, et al. Tracking cells in epithelial acini by light sheet microscopy reveals proximity effects in breast cancer initiation[J]. Elife, 2020, 9: e54066.
[24] GOLDMAN D B. Vignette and exposure calibration and compensation[J]. IEEE transactions on pattern analysis and machine intelligence, 2010, 32(12): 2276-2288.
[25] HUANG Q, GAO W, CAI W. Thresholding technique with adaptive window selection for uneven lighting image[J]. Pattern recognition letters, 2005, 26(6): 801-808.
[26] SAINI R, DUTTA M. Image segmentation for uneven lighting images using adaptive thresh- olding and dynamic window based on incremental window growing approach[J]. International Journal of Computer Applications, 2012, 56(13).
[27] ROHRBACH A. Artifacts resulting from imaging in scattering media: a theoretical prediction [J]. Optics letters, 2009, 34(19): 3041-3043.
[28] DIJKSMAN J A, BRODU N, BEHRINGER R P. Refractive index matched scanning and de- tection of soft particles[J]. Review of Scientific Instruments, 2017, 88(5): 051807.
[29] COUTU D L, SCHROEDER T. Probing cellular processes by long-term live imaging–historic problems and current solutions[J]. Journal of cell science, 2013, 126(17): 3805-3815.
[30] LOEFFLER D, SCHROEDER T. Understanding cell fate control by continuous single-cell quantification[J]. Blood, The Journal of the American Society of Hematology, 2019, 133(13): 1406-1414.
[31] WANG H, ZHU Q, DING L, et al. Scalable volumetric imaging for ultrahigh-speed brain map- ping at synaptic resolution[J]. National Science Review, 2019, 6(5): 982-992.
[32] XU F, SHEN Y, DING L, et al. High-throughput mapping of a whole rhesus monkey brain at micrometer resolution[J]. Nature biotechnology, 2021, 39(12): 1521-1528.
[33] SIEGMAN A E. Lasers university science books[J]. Mill Valley, CA, 1986, 37(208): 169.
[34] MANDEL L, WOLF E. Optical coherence and quantum optics[M]. Cambridge university press, 1995.
[35] SVELTO O, HANNA D C, et al. Principles of lasers: volume 1[M]. Springer, 2010.
[36] DURNIN J. Exact solutions for nondiffracting beams. I. The scalar theory[J]. JOSA A, 1987, 4(4): 651-654.
[37] LIKAR B, MAINTZ J A, VIERGEVER M A, et al. Retrospective shading correction based on entropy minimization.[J]. Journal of Microscopy, 2000, 197(Pt 3): 285-295.
[38] VAN DEN DOEL L, KLEIN A, ELLENBERGER S, et al. Quantitative evaluation of light microscopes based on image processing techniques[J]. Bioimaging, 1998, 6(3): 138-149.
[39] VARGA V S, BOCSI J, SIPOS F, et al. Scanning fluorescent microscopy is an alternative for quantitative fluorescent cell analysis[J]. Cytometry Part A: The Journal of the International Society for Analytical Cytology, 2004, 60(1): 53-62.
[40] YOUNG I T. Shading correction: compensation for illumination and sensor inhomogeneities [J]. Current Protocols in Cytometry, 2000, 14(1): 2-11.
[41] MODEL M A, BURKHARDT J K. A standard for calibration and shading correction of a fluorescence microscope[J]. Cytometry: The Journal of the International Society for Analytical Cytology, 2001, 44(4): 309-316.
[42] MODEL M. Intensity calibration and flat-field correction for fluorescence microscopes[J]. Cur- rent protocols in cytometry, 2014, 68(1): 10-14.
[43] STERNBERG S R. Biomedical image processing[J]. Computer, 1983, 16(01): 22-34.
[44] LEONG F W, BRADY M, MCGEE J O. Correction of uneven illumination (vignetting) in digital microscopy images[J]. Journal of clinical pathology, 2003, 56(8): 619-621.
[45] LIKAR B, MAINTZ J A, VIERGEVER M A, et al. Retrospective shading correction based on entropy minimization.[J]. Journal of Microscopy, 2000, 197(Pt 3): 285-295.
[46] SCHWARZFISCHER M, MARR C, KRUMSIEK J, et al. Efficient fluorescence image nor- malization for time lapse movies[J]. Proc. Microscopic Image Analysis with Applications in Biology, 2011, 5(5).
[47] SINGH S, BRAY M A, JONES T, et al. Pipeline for illumination correction of images for high-throughput microscopy[J]. Journal of microscopy, 2014, 256(3): 231-236.
[48] SMITH K, LI Y, PICCININI F, et al. CIDRE: an illumination-correction method for optical microscopy[J]. Nature methods, 2015, 12(5): 404-406.
[49] PENG T, THORN K, SCHROEDER T, et al. A BaSiC tool for background and shading cor- rection of optical microscopy images[J]. Nature communications, 2017, 8(1): 14836.
[50] KHONINA S N, KAZANSKIY N L, KARPEEV S V, et al. Bessel beam: Significance and applications—A progressive review[J]. Micromachines, 2020, 11(11): 997.
[51] VICENTE O C, CALOZ C. Bessel beams: a unified and extended perspective[J]. Optica, 2021, 8(4): 451-457.
[52] KHONINA S N, KAZANSKIY N L, KARPEEV S V, et al. Bessel beam: Significance and applications—A progressive review[J]. Micromachines, 2020, 11(11): 997.
[53] JIMÉNEZ-GAMBÍN S, JIMÉNEZ N, BENLLOCH J M, et al. Generating Bessel beams with broad depth-of-field by using phase-only acoustic holograms[J]. Scientific reports, 2019, 9(1): 20104.
[54] LUNA-PALACIOS Y Y, LICEA-RODRIGUEZ J, CAMACHO-LOPEZ M D, et al. Multicolor light-sheet microscopy for a large field of view imaging: A comparative study between Bessel and Gaussian light-sheets configurations[J]. Journal of Biophotonics, 2022, 15(6): e202100359.
[55] XIONG B, HAN X, WU J, et al. Improving axial resolution of Bessel beam light-sheet fluores- cence microscopy by photobleaching imprinting[J]. Optics Express, 2020, 28(7): 9464-9476.
[56] FAHRBACH F O, SIMON P, ROHRBACH A. Microscopy with self-reconstructing beams[J]. Nature photonics, 2010, 4(11): 780-785.
[57] GARCÉS-CHÁVEZ V, MCGLOIN D, MELVILLE H, et al. Simultaneous micromanipulation in multiple planes using a self-reconstructing light beam[J]. Nature, 2002, 419(6903): 145-147.
[58] DURNIN J, MICELI JR J, EBERLY J H. Diffraction-free beams[J]. Physical review letters, 1987, 58(15): 1499.
[59] GARCÉS-CHÁVEZ V, MCGLOIN D, MELVILLE H, et al. Simultaneous micromanipulation in multiple planes using a self-reconstructing light beam[J]. Nature, 2002, 419(6903): 145-147.
[60] LIN P Y, HWANG S P L, LEE C H, et al. Two-photon scanned light sheet fluorescence mi- croscopy with axicon imaging for fast volumetric imaging[J]. Journal of Biomedical Optics, 2021, 26(11): 116503-116503.
[61] JIA H, YU X, YANG Y, et al. Axial resolution enhancement of light-sheet microscopy by double scanning of Bessel beam and its complementary beam[J]. Journal of biophotonics, 2019, 12(1): e201800094.
[62] HUISKEN J, STAINIER D Y. Even fluorescence excitation by multidirectional selective plane illumination microscopy (mSPIM)[J]. Optics letters, 2007, 32(17): 2608-2610.
[63] GLASER A K, CHEN Y, YIN C, et al. Multidirectional digital scanned light-sheet microscopy enables uniform fluorescence excitation and contrast-enhanced imaging[J]. Scientific Reports, 2018, 8(1): 13878.
[64] CHANG Y, YAN L, WU T, et al. Remote sensing image stripe noise removal: From image decomposition perspective[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 54(12): 7018-7031.
[65] MÜNCH B, TRTIK P, MARONE F, et al. Stripe and ring artifact removal with combined wavelet—Fourier filtering[J]. Optics express, 2009, 17(10): 8567-8591.
[66] FEHRENBACH J, WEISS P, LORENZO C. Variational algorithms to remove stationary noise: applications to microscopy imaging[J]. IEEE transactions on image processing, 2012, 21(10): 4420-4430.
[67] ESCANDE P, WEISS P, ZHANG W. A variational model for multiplicative structured noise removal[J]. Journal of Mathematical Imaging and Vision, 2017, 57: 43-55.
[68] LIANG X, ZANG Y, DONG D, et al. Stripe artifact elimination based on nonsubsampled contourlet transform for light sheet fluorescence microscopy[J]. Journal of Biomedical Optics, 2016, 21(10): 106005-106005.
[69] DA CUNHA A L, ZHOU J, DO M N. The nonsubsampled contourlet transform: theory, design, and applications[J]. IEEE transactions on image processing, 2006, 15(10): 3089-3101.
[70] WEI Z, WU X, TONG W, et al. Elimination of stripe artifacts in light sheet fluorescence mi- croscopy using an attention-based residual neural network[J]. Biomedical Optics Express, 2022, 13(3): 1292-1311.
[71] STRINGER C, WANG T, MICHAELOS M, et al. Cellpose: a generalist algorithm for cellular segmentation[J]. Nature methods, 2021, 18(1): 100-106.
[72] WANG Z. Fast algorithms for the discrete W transform and for the discrete Fourier transform [J]. IEEE Transactions on Acoustics, Speech, and Signal Processing, 1984, 32(4): 803-816.
[73] KHAYAM S A. The discrete cosine transform (DCT): theory and application[J]. Michigan State University, 2003, 114(1): 31.
[74] SAZLI M H. A brief review of feed-forward neural networks[J]. Communications Faculty of Sciences University of Ankara Series A2-A3 Physical Sciences and Engineering, 2006, 50(01).
[75] SALEHINEJAD H, SANKAR S, BARFETT J, et al. Recent advances in recurrent neural net- works[A]. 2017.
[76] O’SHEA K, NASH R. An introduction to convolutional neural networks[A]. 2015.
[77] KINGMA D P, BA J A, ADAM J. A method for stochastic optimization. arXiv 2014: volume 106[A]. 2020.
[78] XIAO L, FANG C, ZHU L, et al. Deep learning-enabled efficient image restoration for 3D microscopy of turbid biological specimens.[J]. Optics Express, 2020, 28(20): 30234-30247.
[79] GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial networks [J]. Communications of the ACM, 2020, 63(11): 139-144.
[80] MIRZA M, OSINDERO S. Conditional generative adversarial nets[A]. 2014.
[81] RADFORD A, METZ L, CHINTALA S. Unsupervised representation learning with deep con- volutional generative adversarial networks[A]. 2015.
[82] ISOLA P, ZHU J Y, ZHOU T, et al. Image-to-image translation with conditional adversarial networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 1125-1134.
[83] WANG T C, LIU M Y, ZHU J Y, et al. High-resolution image synthesis and semantic manip- ulation with conditional gans[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 8798-8807.
[84] WOO S, PARK J, LEE J Y, et al. Cbam: Convolutional block attention module[C]//Proceedings of the European conference on computer vision (ECCV). 2018: 3-19.
[85] MAO X, LI Q, XIE H, et al. Least squares generative adversarial networks[C]//Proceedings of the IEEE international conference on computer vision. 2017: 2794-2802.