CASIA OpenIR  > 毕业生  > 博士学位论文
生物自发荧光三维断层成像方法研究
Alternative TitleResearch on Three-Dimensional Reconstruction Methods for Bioluminescence Tomography
刘凯
Subtype工学博士
Thesis Advisor田捷
2011-05-30
Degree Grantor中国科学院研究生院
Place of Conferral中国科学院自动化研究所
Degree Discipline模式识别与智能系统
Keyword分子影像 光学成像 生物自发荧光断层成像 前向数理模型 重建算法 逆问题 高阶近似模型 多模态融合 Molecular Imaging Optical Imaging Bioluminescence Tomography Forward Math-physical Model Reconstruction Algorithm Inverse Problem Higher-order Approximation Model Multimodality Fusion
Abstract近十年来,光学成像等分子影像模态蓬勃发展,大大推动着医学影像技术的进步。通过分子标记的特异性成像可从功能、代谢上拓展原有传统成像模态的成像能力。分子影像将不仅能使得医生看到肿瘤在体内的位置,而且还可将影响肿瘤行为以及肿瘤对药物治疗反应的特定分子的表达和活动以及生物过程可视化。这些信息都将对癌症诊断、个性化治疗、药物研发以及理解癌症如何发生发展提供有效信息。生物自发光断层成像是光学分子影像的一种重要模态,通过探针特异性地标记靶点从而对病灶分布进行定位和量化。 经过几年的研究,生物自发光断层成像的成像理论已初步建立起来。然而由于该成像从物理本质上是要运用成像目标表面上获取到的有限信息来估计内部生物自发光光源的分布,而且光在在动物组织中传播时要经历多次的散射和吸收更是增大了该问题的难度,是一个典型的病态问题。因此,如何发展新型的成像模型和算法,降低对苛刻成像条件的依赖,提高图像重建的质量、效率、鲁棒性等都是国际上一直在不断探索的新问题。本文也正是在此背景下开展了生物自发光断层成像技术的研究。本文从断层成像的各个技术难点和实用化需求着手,重点研究了成像技术中求解逆问题的重建算法和描述光在动物组织中传播的前向数理模型:在逆问题方面深入研究,研发多种重建技术提高成像质量与效率;在前项问题方面引入了一种新型的高阶近似模型,从数理模型上提高模型描述的精确度进而提高成像质量。此外,对多模态融合技术也做了有益的探索。主要研究内容包括: 1.针对现有算法不能有效提取重建光源轮廓的问题,提出了一种基于多光谱的水平集生物自发光断层成像算法。通过水平集函数控制光源的定位,通过在每个水平上的值来控制光源的能量密度,在迭代中重建光源的位置、强度的同时,有效利用水平集的优势,重建光源的大体形状。在数值验证中,在不同的初始值和噪声水平下,本方法都能重建得到可靠的结果。小鼠仿体实验进一步证明了本方法可以用于生物医学相关成像的应用。真实实验更加全面地验证了本算法在实际应用中的有效性。 2.针对一些医学应用如检测淋巴癌需要全身成像的现实需,提出了一种基于广义图割的定位生物自发光光源的重建算法。传统的图割算法只能解决submodular的问题,但是由于在光学成像中含有的supermodular项。因此使用了二次伪布尔最优化技术将图割技术广义化,使得图像重建的目标函数改造为可解决的形式。在数值验证上,首先与基于梯度的传统方法对比中验证了本算法在未融合光源位置的先验信息条件下可进行全域重建。在匀质、非匀质结构信息条件下以及光学参数误差存在下的数值验证从不同角度测试了本算法的成像效果。然后又通过一个真实动物实验,进一步验证了本算法在具体肿瘤机理研究和药物研发中的应用潜力。 3.针对现有算法在实际断层图像重建中不够鲁棒的问,提出了一种鲁棒高效的基于动态稀疏正则化的全局非精确牛顿法。通过一个权重二次范数动态模拟稀疏正则化项来充分利用光源稀疏分布的先验信息。含有一个回溯策略的非精确牛顿法则可在多个极小值中搜索到全局最优化的...
Other AbstractOver the last decade, molecular imaging including optical imaging, which has undergone explosive growth, has tremendously promoted the progress of medical imaging. Based on specifically targeted imaging agents, molecular imaging is expected to broadly expand the capabilities of conventional imaging methods. Molecular imaging will allow scientists and clinicians to not only see where a tumor is located in the body, but also to visualize the activity and expression of specific molecules and biological processes that influence tumor behavior and/or response to therapy. This information is expected to have a major impact on cancer detection, drug development, individualized treatment, as well as understanding of how cancer arises. Bioluminescence tomography is one of important modalities in molecular imaging, and it can localize and quantify the cancer distribution using specifically targeted agents. After several years of continuous research, the imaging theory of bioluminescence tomography has been preliminarily established. Nevertheless, it is known that the inverse problem of tomographic imaging is an ill-posed problem due to the fact that only some limited information can be measured from the boundary of animals to estimate the internal bioluminescent source distribution. The difficulty of the problem is exacerbated as photon has to undergo multiple scattering and absorption when transporting through tissues. Therefore, the international community has been constantly exploring how to reduce the dependency with the rigorous imaging requirements, how to improve the imaging quality, efficiency, and robustness of the reconstruction methods, etc. Based on the background above, related research for bioluminescence tomography has also been carried on in this thesis. According to the difficulty and practical utilization, this thesis focuses on the reconstruction methods to solve the inverse problem and forward problem to accurately describe photon transporting through tissues. In the aspect of study on inverse problem, reconstruction methods have been developed to improve imaging quality and efficiency; in the aspect of study on forward problem, a high-order approximation model has been introduced, and it improved the accuracy for modeling the transporting process to maintain high level imaging quality. Moreover, the multimodality fusion method has been also explored beneficially. The main research work includes: 1.Since the existing methods can not effecti...
shelfnumXWLW1580
Other Identifier200818014628046
Language中文
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/6373
Collection毕业生_博士学位论文
Recommended Citation
GB/T 7714
刘凯. 生物自发荧光三维断层成像方法研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2011.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20081801462804(13768KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[刘凯]'s Articles
Baidu academic
Similar articles in Baidu academic
[刘凯]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[刘凯]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.