自发荧光断层成像逆向问题研究
文献类型：学位论文
| 作者 | 吕玉杰 |
| 学位类别 | 工学博士 |
| 答辩日期 | 2007-06-20 |
| 授予单位 | 中国科学院研究生院 |
| 授予地点 | 中国科学院自动化研究所 |
| 导师 | 田捷 |
| 关键词 | 光学分子影像 自发荧光断层成像 自适应有限元方法 蒙特卡洛方法 重建算法 系统构建 Optical molecular imaging bioluminescence tomography (BLT) adaptive finite element methods Monte Carlo methods reconstruction algorithm system installation |
| 其他题名 | Research of Inverse Problem in Bioluminescence Tomography (BLT) |
| 学位专业 | 模式识别与智能系统 |
| 中文摘要 | 为发展下一代多模态多光谱分子影像系统，本文的工作从成像理论、重建算法以及系统验证与构建等多个方面对自发荧光断层成像技术进行了深入研究，主要内容包括： 1.发展了一种适合于在复杂生物组织中进行光子传播仿真的自适应有限元方法以及进一步完善了在体生物光学前向问题仿真平台（MOSE）。该自适应有限元方法利用易于表达复杂几何形体的四面体单元作为区域网格剖分的基本单元和利用后验误差估计的方法来选择部分网格单元进行局部细分。该方法还进一步考虑固有的剖分网格序列发展了BPX预处理子来加速仿真求解，从而在提高仿真精度的同时也改善了仿真的速度。另外，通过验证MOSE 平台对非匀质仿体仿真的正确性以及添加多光谱的仿真算法，进一步发展了MOSE平台的功能。 2.提出了一种多水平自适应有限元光源重建算法。该算法利用自适应有限元方法结合了先验可行光源区域策略来解决光源重建中存在的重建质量与区域离散粗细之间的矛盾，并融合多水平的方法进一步改善了重建质量，提高了数值稳定性和有效性。在仿真实验中，通过考虑inverse crime问题，利用蒙特卡洛的仿真测量数据验证了该算法在单源以及多源下的重建能力，并进一步获得了毫米域的光源重建分辨率。 3.提出了一种多光谱多尺度光源重建算法。该算法在前一种算法的基础上融合了多光谱测量信息进而发展了后验可行光源区域的策略，结合局部的网格细分实现了光源的多尺度重建。通过匀质仿体和microMRI 获取的整体小鼠模型，评估了该算法对不同深度光源重建问题的有效性以及对光学特性参数误差的容忍性。 4.实验验证了多模态融合对自发荧光断层成像的必要性。在自发荧光信号采集系统中，通过对非匀质仿小鼠仿体进行多角度非接触式的探测，利用多水平自适应有限元重建方法验证了解剖结构信息和光学特性参数信息对光源重建的必要性，为下一代自发荧光断层成像原型系统的构建提供了较好的技术验证。 |
| 英文摘要 | For the next generation optical molecular imaging system, deeply research need to be performed at its various aspects, such as photon propagation, source reconstruction and so on. In term of imaging theory and algorithm, and system verification and installation of bioluminescence tomography, the main work of this study is as follows: 1.Develop an adaptive finite element method appropriate for photon propagation in complex biological tissues and further improve the Molecular Optical Simulation Environment (MOSE). The developed adaptive finite element method uses tetrahedron suitable for description of complex geometry as basic element and performs local mesh refinement based on a posteriori error estimation technique. Considering the intrinsic triangulation sequence obtained by adaptive mesh refinement, BPX preconditioner is also used to significantly accelerate iterative procedure in solving linear equation. The use of these strategies above make us not only obtain high numerical precision but also improve simulation speed. In addition, the MOSE platform is improved through validating the heterogeneous phantom-based simulation and adding multispectral simulation algorithm. 2.Propose a multilevel adaptive finite element algorithm for bioluminescent source reconstruction. Through the combination of the developed adaptive finite element method and a priori permissible source region, the algorithm resolves the contradiction between reconstruction quality and speed and the fixed discretization degree to the given domain in BLT. The multilevel strategy is also fused into the algorithm which enhances numerical stability and robustness. In simulation experiments, in term of inverse crime problem, the ability of the algorithm is verified in reconstructing single and multiple bioluminescent sources. Furthermore, a spatial resolution in the millimeter domain is obtained. 3.Propose a multispectral multiscale source reconstruction algorithm. Based on the multilevel adaptive finite element algorithm, the algorithm considers multispectral detection information to develop a posteriori permissible source region strategy. The algorithm further combines local mesh refinement to realize multiscale source reconstruction. Using the homogeneous and microMRI-based whole-body mouse phantoms, the algorithm represents its ability of reconstructing bioluminescent source in different depths and tolerating optical property errors. 4.Experimentally verify the necessity of multimodality imaging fusion to bioluminescent tomography. In a bioluminescent signal collection system, a heterogeneous mouse-like phantom is detected by using the multiview noncontact mode. The BLT reconstruction with the multilevel adaptive finite element algorithm demonstrates the necessity of anatomical information and tissue optical property. This provides a preferable validation for installing the next generation BLT prototype. |
| 语种 | 中文 |
| 其他标识符 | 200418014628052 |
| 源URL | [http://ir.ia.ac.cn/handle/173211/6030]  |
| 专题 | 毕业生_博士学位论文 |
| 推荐引用方式 GB/T 7714 | 吕玉杰. 自发荧光断层成像逆向问题研究[D]. 中国科学院自动化研究所. 中国科学院研究生院. 2007. |
入库方式： OAI收割
来源：自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。