CASIA OpenIR  > 毕业生  > 博士学位论文
基于SP3 的契伦科夫荧光断层成像重建方法研究
其他题名Reconstruction Methods for Cerenkov Luminescence Tomography via SP3
钟江宏
2012-05-30
学位类型工学博士
中文摘要在体监测细胞分子水平的分子探针与生物参数,是分子影像相比于传统医学影像技术的显著进步。随着研究与应用的纵深拓展,微量探针的多功能检测与生物参数的多角度跟踪,成为分子影像技术的发展趋势。它的技术创新在多模态融合方法。以分子影像为契机和纲要,研究多模态实质性融合重建方法,是以细胞分子、功能代谢、解剖结构多源信息实现疾病的早期诊断、精确治疗以及药物创制的在体评判与动态监测技术的新突破。 契伦科夫荧光断层成像(Cerenkov luminescence tomography, CLT)是基于契伦科夫效应的多模态分子影像,在功能与结构信息融合的同时,不仅继承了分子核医学的高灵敏和光学分子影像的低成本特征,而且具备临床应用价值。CLT的基本流程是光学分子影像设备多角度采集医学同位素诱发的契伦科夫光子,耦合解剖结构信息重建三维断层图像。鉴于成像设备和探针可以借助于市面成熟产品,断层图像重建就成为CLT的研究难点与应用前提。从数学建模与计算角度分析,契伦科夫荧光穿透力的微弱性、微量放射性药物在体分布的全域稀疏性和局部聚集性,增强了CLT重建高维泛函线性化复杂度及线性空间的不适性,严重影响近似解的精度和收敛性。提出并证实CLT的全域快速重建方法是本文的研究所在,主要包括以下三部分内容: 高灵敏是同位素示踪技术特征,奠定分子核医学应用基础;低成本是光学成像系统硬件特征,降低光学分子影像研究门槛;特异性是多模态信息融合与计算特征,提高影像信息准确度和实用性。 基于SP3<下标!>方法的前向重建模型研究。契伦科夫光子在生物组织中的传播规律及其数值模型是CLT准确重建基础。设计18<上标!>F-FDG在体成像实验,运用扩散近似模型和L2正则化方法重建断层图像,挖掘契伦科夫荧光特点和CLT重建难点;针对契伦科夫荧光与组织相互作用规律,建立改进型的基于三阶简化球谐近似(The third-order simplified spherical harmonics approximation, SP3<下标!>)的前向模型;参比蒙特卡洛方法,采用复杂组织单双光源数值仿真实验,定量定性验证前向模型解的正确性和精度。 基于Lp正则化的CLT全域重建研究。全域重建的难点是更高维数据空间的不适定问题快速求解。构造泛函拓扑度量是恢复病态问题适定性和收敛性的重要方法,例如添加Lp正则化信息并改变p值使得度量不同。引用基于内点的L1<下标!>正则化方法,结合预处理共轭梯度算法,初步实现大规模数据求解任务;平衡正则化解的稀疏性和光滑性,提出弹性网正则化(Elastic-net regularization)方法,通过数值仿真、真实仿体实验和在体应用对比论证其对CLT重建精度与强度的影响;针对欠定反演泛函空间解的稀疏性,设计非凸性、非一致连续、无偏L1/2<下标!>正则化算子,耦合几何尺度变换方法提高正则化解逼近速度,通过仿真与在体实验说明其计算效能。 基于AMP的稀疏性可调的重建研究。医学同位素分布的稀疏性客观存在但未知,且CLT反演重建测得数据不再符合香农条件。针对欠定反问题解空间维数适定问题,提出预定正则化解稀疏性的重建策略,应用近似信息传递(Approximate message-pas...
英文摘要In vivo visualization of molecular imaging agents and biological parameters at the molecular and cellular levels is the significant progress of molecular imaging in biomedical imaging technology, compared to traditional modes. With the research and application development, multi-function testing of micro-dose probes and multi-angle track of biological parameters become the cutting-edge science and technology trends of molecular imaging. This research hopes to contribute to the advancement of molecular imaging into a new stage of the multi-modal integration. Multi-modal molecular imaging will provide molecular, functional and anatomical information for preclinical and clinical services, such as early diagnosis and precise treatment of diseases, in vivo evaluation and dynamic monitoring of drugs. Cerenkov luminescence tomography (CLT) is a multi-modal molecular imaging based on the Cerenkov effect, which not only inherits the characteristics of high sensitivity in nuclear medicine and low cost in optical molecular imaging, but also has even more potential in clinical measurement. The basic process of CLT includes multi-angle measurement of medical isotopes by optical molecular imaging device, coupling anatomical structure, and 3D tomography image reconstruction. In view of the imaging device and probe from prototype or mature products, 3D reconstruction becomes the key issue of CLT research and application. From the view of mathematical and computational modeling, the weakness of Cerenkov light penetration, the trace of radioactive drugs in the body with the global sparsity and local aggregation will enhance the ill-posedness of inverse problems, seriously affecting the accuracy and convergence of the approximate solution. Therefore, this thesis focused on treating the ill-conditioned problem of fast reconstruction for whole-body CLT, of which the main contributions are listed as follows: 1. Research on the forward model of CLT through the third-order simplified spherical harmonics approximation (SP3<下标!>). Cerenkov photon propagation in biological tissue and its numerical model is the basis of accurate reconstruction. Designing in vivo 18<上标!>F-FDG experiments, we studied the characteristics of Cerenkov radiation and inverse reconstruction by use of diffusion approximation model and L2<下标!> regularization method. The establishment of the improved SP3<下标!> method was so imperative as to model Cherenkov luminescence e...
关键词多模态分子影像 契伦科夫荧光断层成像 球谐近似 Lp 正则化 近似信息传递 18f-fdg Multimodality Molecular Imaging Cerenkov Luminescence Tomography Spherical Harmonics Approximation Lp Regularization Approximate Message-passing 18f-fdg
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/6448
专题毕业生_博士学位论文
推荐引用方式
GB/T 7714
钟江宏. 基于SP3 的契伦科夫荧光断层成像重建方法研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2012.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
CASIA_20091801462807(13634KB) 暂不开放CC BY-NC-SA
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[钟江宏]的文章
百度学术
百度学术中相似的文章
[钟江宏]的文章
必应学术
必应学术中相似的文章
[钟江宏]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

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。