基于序贯贝叶斯分析的康普顿成像快速识别方法研究

Compton imaging is an advanced passive gamma ray imaging technology that can realize spatial localization and contour imaging of radioactive sources. It has broad application prospects in the fields of public safety, medical imaging, astrophysics and arms control verification. At present, relevant research in the world has a common problem of low imaging efficiency and long imaging time, which makes the practical application of Compton imaging technology limited. In recent years, sequential Bayesian analysis method, which is emerging in the field of radionuclide identification, combined with Bayes' rule and sequential probability ratio test theory, has been able to make real-time judgments rapidly on the existence of radionuclides. In essence, Compton imaging technology and radionuclide identification have certain similarities, which can be taken as abnormal discrimination of specific attributes. This project, titled of “Rapid identification method in Compton imaging based on the sequential Bayesian analysis”, aims to introduce sequential Bayesian analysis method into Compton imaging technology, to study the intrinsic coupling relationship between sequential Bayesian analysis method and Compton imaging theory, and to develop a rapid identification method for the specific target shapes, based on which the bottleneck of that the existing Compton imaging reconstruction methods are difficult to balance imaging precision and imaging time will be broken through. Research conclusions of this project must be instructive and significant for the design of Compton imaging system and for the improvement of image reconstruction method, provide a theoretical basis and technical supports for promoting the practical application of Compton imaging technology.

康普顿成像是一种先进的被动式伽马射线成像技术，可以实现对放射源的空间定位和轮廓成像，在公共安全、医学成像、天体物理和军控核查等领域具有广阔的应用前景。目前国际上相关研究普遍存在成像效率低和成像时间长的问题，使得康普顿成像技术的实际应用受限。而近年来在放射性核素识别领域兴起的序贯贝叶斯分析方法，结合贝叶斯法则和序贯概率比检验理论，已经能够对放射性核素的存在性做出快速实时判断。从本质上讲，康普顿成像技术与放射性核素识别具有一定的相似性，均可理解为特定属性的异常判别。本课题拟将序贯贝叶斯分析方法引入康普顿成像技术，研究序贯贝叶斯分析方法与康普顿成像理论的内在耦合关系，建立一种针对特定目标物形状的快速识别方法，突破现有康普顿图像重建算法难以兼顾成像精度和成像时间的瓶颈问题。本课题的研究结论对于康普顿成像系统设计和图像重建算法改进具有指导意义，可为推动康普顿成像技术的实际应用提供理论依据和技术支持。

康普顿成像是一种先进的被动式伽马射线成像技术，可以实现对放射源的空间定位和轮廓成像，在公共安全、医学成像、天体物理和军控核查等领域具有广阔的应用前景。目前国际上康普顿成像相关研究普遍存在成像效率低和成像时间长的问题，使得康普顿成像技术的实际应用受限。而近年来在放射性核素识别领域兴起的序贯贝叶斯分析方法，结合贝叶斯法则和序贯概率比检验理论，已经能够对放射性核素的存在性做出快速实时判断。从本质上讲，康普顿成像技术与放射性核素识别具有一定的相似性，均可理解为特定属性的异常判别。.本项目利用Geant4蒙特卡洛数值模拟程序和MATLAB数据分析软件，以双层CZT像素阵列探测器为基础，构建了一系列基于序贯贝叶斯分析的康普顿成像系统物理模型。首次将序贯贝叶斯分析方法引入康普顿成像技术，建立康普顿成像与序贯贝叶斯分析的耦合理论，拓展了序贯贝叶斯分析方法在辐射成像领域的应用实例。结合数值模拟、理论分析及实验验证，提出针对特殊目标物形状的快速识别方法和针对任意形状放射源的快速重建算法，从一定程度上解决了现有算法难以兼顾成像精度和成像时间的瓶颈问题，有助于进一步推动康普顿成像技术的实际应用。