【作者】 邓宁；

【导师】 段会龙；

【作者基本信息】 浙江大学 ， 生物医学工程， 2007， 博士

【摘要】 生物芯片技术在最近十几年发展迅速，已经广泛应用于基因表达、功能基因组、蛋白质组等前沿领域。尤其在近几年，所涉及的领域从研究型场合迅速扩展至应用型场合，在临床检验、疾病诊断、药物筛选等方面发挥了越来越重要的作用。用生物芯片进行检测和分析的一大特点是需要处理大量数据，实现多通道、高通量、全自动化的数据处理。生物芯片分析是生物芯片技术的研究热点，包括靶点阵列定位、靶点分割、数据提取和表达分析等子过程。它不是单一的算法或方法，而是涵盖生物芯片分析全过程的系列算法的集合。目前生物芯片分析方法的研究更多的局限于某个分析环节。要真正实现高效、全自动的生物芯片分析，必须从两方面着手：首先，需要对分析的全过程进行研究，解决每个分析环节中存在的问题；其次，需要在各个分析环节之间建立有效的联系。随着生物芯片技术逐步应用于临床领域，能否在各个分析环节形成公认有效的分析算法，如何建立连续的分析流程，提高分析的自动化程度将显得越来越重要。论文对生物芯片分析方法中所包含的各个分析子过程进行了深入的研究，在每个分析子过程建立了高性能的分析算法，并进行了实验分析验证。主要的工作包括：提出了基于投影的生物芯片网格阵列自动定位算法，通过引入网格阵列的分布参数，实现对靶点阵列的自动定位。利用图像投影的能量谱密度，实现对倾斜图像的自动校正，并通过多尺度快速校正算法进行性能优化。在分析并比较了现有生物芯片图像分割方法的基础上，采用结合邻域搜索的自适应圆形分割算法，实现了对生物芯片图像的分割。以临床检验生物芯片为主要研究对象，建立了面向临床应用的差异表达分析模型，关键算法包括：基于变异系数逼近的数据筛选策略，可调控的低表达水平信号分析模型等。该分析模型在丙型肝炎病毒检测芯片分析中进行了验证，提高了检测结果的一致率，降低了假阳性、假阴性率。面向临床应用，开展了生物芯片自动分析平台的研发工作，包括：激光共聚焦荧光分析仪的研制，生物芯片自动分析工作站，生物芯片数据库的构建等。分别用80例丙型肝炎病毒检测芯片标准品、256例乙型肝炎病毒前C区变异基因芯片临床样本和63例过敏原检测芯片临床样本进行了实验和分析，验证了论文方法的准确性和可靠性。实验和分析表明，论文所提出的一系列分析算法和方法能对微阵列生物芯片进行全自动分析，通过算法间的有效联系，实现了网格定位、靶点分割和表达分析等过程的自动串联：临床应用的结果表明，论文所建立的生物芯片自动分析平台，为生物芯片研制、开发和应用提供了有效的方法和技术。更多还原

【Abstract】 With the advance in recent 10 to 20 years, microarray has been widely developed in gene expression, functional genome, proteome and other research areas. Especially in recent years, it has been involved from the research fields to the clinical fields with the more and more important effect in clinical inspection, diagnosis, drug filtration etc. An issue of microarray experiment is to process large-scale of data with multi-channel, high through pass and automation.Microarray analysis is an important aspect of microarray technology. It has several processes including gddding, segmentation, data extraction and expression. Instead of a certain algorithm or method, it is a set of methods which cover the whole procedure of microarray experiment, but most of the current research works focus on certain analysis step. For the realization of high performance and automatic microarray analysis, the study of whole analysis procedure should be considered, not only solve the issues in each step, but also research on the relationship between them. With micorarray technology’s wide implementation in clinical area, the breakthrough of microarray analysis becomes a desire.This thesis focuses on the methodology study of microarray analysis. A set of algorithms and methods which relate to each step of micorarray experiment were presented and validated by a great deal of experiments. The main contributions of this thesis are followings:An automatic gridding algorithm of biochip image based on projection is presented.For microarray image rotated correction, an automatic and power spectrum-based rotated correction algorithm is presented. For improving the performance, the optimization with multi-scale searching method is introduced.Based on the comparison of existed microarray image segmentation methods, the adaptive circle segmentation algorithm combined with the neighborhood searching algorithm is realized.The microarray data differential expression model is realized for clinical inspection. The key algorithms include: data filtration strategy based on coefficient variant, adjustable model for low level expression sample etc. This model is implemented for hepatitis C virus detection microarray. The analysis result showed a high consistence rate.Aiming at clinical implementation, a microarray analysis platform is researched and developed, including the development of laser confocal microarray scanner, microarray automatic analysis workstation, microarray database etc.A great deal of experiments were done to evaluate the performance of our algorithms and methods, including 80 samples of hepatitis C virus detection microarray, 256 samples of mutation detection microarray for hepatitis B virus the precore region, 63 samples of allergen detection microarray.Experiments and analysis show that the algorithms and methods presented in this thesis can be used for automatic microarray analysis. With an effective interaction between every analyzing step, gridding, segmentation and data expression can be processed in series. The result of clinical implementation also shows that the microarray analysis platform provides effective and feasible methods and technologies for microarray research, development and implementation.更多还原