【作者】 翁嘉文；

【导师】 钟金钢；

【作者基本信息】 暨南大学 ， 生物医学信息技术， 2012， 博士

【摘要】 生命科学研究的突破在很大程度上依赖于各种新颖的、性能高的分析测试仪器。显微成像设备是生命科学研究中最常见的分析测试设备。传统的显微镜已经远不能满足目前科学研究的要求。发展新的显微成像分析技术，对活细胞等微小生物样品实现无扰、动态、定量的显微分析是目前生命科学研究的迫切需求。该技术关系到疾病发生机制的解析、医疗效果评价、药物筛选等重大科学问题。随着计算机和高分辨率CCD摄像机的发展，以及数字图像处理技术的进步，结合数字全息成像技术以及光学条纹分析技术，逐渐出现了基于复振幅位相定量分析的数字全息显微成像技术。该技术采用CCD摄像机等电子成像器件代替全息干版来记录全息图，可连续记录运动物体的各个瞬间过程，实现连续在线全息记录，并将记录的全息图存入计算机；其再现过程是对全息图进行数值再现，可定量分析样品的幅度和位相信息，为样品的定量检测分析带来了极大的方便。目前数值再现重构技术，主要是以菲涅尔衍射积分公式为基础的菲涅耳算法、卷积算法以及基于角谱传播理论的角谱法。但上述几种方法都借助了傅里叶变换算法来获得全息图的频谱，并基于空间滤波技术设计滤波器，在频域中对全息图进行滤波，以消除零级像以及孪生像，提取+1级频谱重构再现像。但当全息图中含有噪音或者寄生衍射条纹时，+1级频谱的边界将会变得模糊不清，因此难以精确地确定滤波窗口的形状及带宽。另外对于不同物体所形成的全息图频谱分布形状并不相同，因而在滤波过程中对于滤波窗口形状，以及大小的选择具有较大的任意性。根据每幅不同的全息图进行人工定义滤波器的方法在一定程度上解决了上述的不足，但对于动态全息分析而言，将有大量的全息图被记录和分析，人工定义滤波器的方法过于耗时，是不现实的。另一方面由于再现像质量与滤波窗口的选择紧密相关，人工定义滤波窗口没有统一的准则。因此，针对动态全息图的特点，对前后两幅全息图再现位相进行比较时，如果空间滤波不当的话，很可能导致前后两幅全息图的比较结果不可靠。因此迫切需要寻找具有统一准则的数值再现算法实现动态数字全息分析。本文为探索一种提高数字全息定量相衬显微动态分析技术，针对目前常用的数值再现算法不适用于动态分析的缺陷，提出一种基于小波变换的数值再现算法，无需空间滤波，避免或减轻动态全息图再现结果比较的不可靠性，实现对活细胞等位相物体的无扰、定量、动态分析。研究内容主要包括理论研究，计算机数值模拟研究以及实验研究。理论研究主要包括以下几方面：（一）基于空间滤波技术，提出自适应滤波器的自动设计算法。（二）提出基于一维以及二维Gabor小波变换的数值再现算法用于动态分析，包括最佳母小波函数的选择、小波变换最佳尺度参数的选取原则、以及小波变换脊的精确提取算法等。（三）考虑在超快动态数字全息中，由于CCD反应速度的局限，需利用角复用技术，提出针对角复用全息图的小波变换物光波数字重建技术。计算机数值模拟研究以及实验研究方面：首先利用数值方法进行模拟，检验上述再现算法的可靠性，并基于Matlab语言开发相关的程序；然后搭建全息实验光路，以洋葱细胞以及生物细胞为实验样品，采用CCD摄像机采集样品的全息条纹，进行数值再现，从而获得生物样品的定量位相分布信息；最后，以一个活动幼虫为研究对象，实现数字全息显微成像的动态分析。本论文的特色与创新之处在于首次在国内外提出应用具有多尺度细化分析的功能的Gabor小波变换构造数字全息再现核心算法，其关键技术在于通过小波变换脊对样品信号基频信息的提取以消除零级像与孪生像。该技术为活细胞等生物样品的分析提供了一种崭新的方法，对探索生命过程，促进生物医学研究实现新的突破，实现对活体细胞的动态检测，具有重要的科学意义。更多还原

【Abstract】 To a great degree, the breakthrough in the field of the research of the life sciences dependon varies novel and powerful analysis measurement instruments. The microscopyapparatus is the most familiar employed. However, the conventional microscopyapparatus can not meet the development trend of the scientific research. Therefore, forthe active demand of the research of the life sciences, it is necessary to grope for a novelmicroscopical imaging technique that can achieve real-time, non-interference andquantitative analysis. This immediately concerns some important scientific problems,such as the exploration of mechanism of disease, the evaluation of the medical results,the drug screening and so on.With the rapid development of the computer and the charge coupled device (CCD)camera sensor technology, great progress had been made in technology of digital imageprocessing. At the same time, in combination with the holography and the optical fringeanalysis method, the digital holographic microscopy, a noninvasive contrast imagingtechnique allowing quantitative visualization of living cells, is developed. A CCD camerais employed to record a hologram onto a computer and numerical methods aresubsequently applied to reconstruct the holographic image to enable direct access to bothphase and amplitude information of the object wave in digital holography. It offers asignificant advantage for dynamic real-time analysis.The most popular numerical reconstruction methods for digital holography includethe well-known Fresnel diffraction integral method, the angular spectrum method and theconvolution-based method. In order to filter out the zero-order term, the twin image termand the parasitic interferences, the process of the spatial filtering must be carried out inthese methods. When some noises and parasitic interferences are introduced into thehologram, the spectrum of the virtual image would be disturbed by some other spectrum.It brings difficulties to define the spatial filter because of the blurry boundary andnon-regular distribution of the spectrum. Therefore, the quality of the reconstructedimage is mainly limited by the process of the spatial filtering. For the analysis ofdifferent object waves with respective spectrum, manual spatial filters with different pass-band are proposed. However, it shows that different results would be obtainedaccording to the spatial filters at different standard in the experiments, and it still bringsdifficulties to define a proper spatial filter at the uniform standard. On the other hand, forthe dynamic analysis, lots of holograms are recorded. Defining different manual spatialfilters would consume plenty of time for dynamic analysis. Thereby, the process of thespatial filtering limits the application of the digital holography for the dynamic andautomatic analysis. Therefore, it is necessary to grope for a numerical reconstructiontechnique that can be performed automatically at the uniform standard.In order to overcome the defect of the most popular numerical reconstructionmethods for the dynamic analysis, a numerical reconstruction technique of digitalholography by means of wavelet transform is described. Appling the wavelet transformto digital holography, the object wave can be reconstructed by calculating the waveletcoefficients of the hologram at the ridge or the peak of the wavelet coefficientsautomatically. At the same time the effect of the zero-order diffraction image and thetwin-image are eliminated without the spatial filtering. The theory, the results of thesimulations and experiments are demonstrated in detail in this paper.The theory mainly includes the following aspects:(1) Based on the processing ofthe spatial filtering, an automatic spatial filtering to obtain the virtual image term indigital holographic microscopy is presented.(2) A numerical reconstruction technique fordigital holography by means of the one-dimensional and two-dimensional Gabor wavelettransform is presented to achieve the dynamic and automatic analysis. The researchincludes the choice of the optimum mother wavelet, the decision of the values of thescale parameter and the accurate determination of the ridge of the Gabor wavelettransform.(3) A numerical reconstruction technique employing the de-multiplexing bymeans of the Gabor wavelet transform in digital holography is described.The simulations and experiments researches are mainly include the followingaspects: first, the numerical simulations are performed to demonstrate the validity of theaforementioned methods, and develop the programs and the application system byMatlab. Second, employ the apparatus analogous to a Mach-Zehnder interferometer forexperiments. Employ a CCD camera to recorder the holograms of the onion specimen and gastric cancer cells and perform the numerical reconstruction method. Finally, asequence of holograms of a grub is recorded for the dynamic and automatic analysis.The creative idea of this paper is that the Gabor wavelet transform, a tool excellingin multiresolution and localization in the time-or space-frequency domains, is applied tothe digital holography. The object wave can be reconstructed by calculating the waveletcoefficients of the hologram at the ridge or the peak of the wavelet coefficientsautomatically without the processing of the spatial filtering. This technique provides anovel method for the analysis of the biological samples, special for the living cells. It isof great significance in the development of the research of the life sciences and thedynamic analysis.更多还原