基于符合光子物理模型的定量PET算法

Quantification inaccuracy is the bottleneck that limits the application of Positron emission tomography (PET) in early diagnosis, treatment evaluation and drug development of disease like cancer and dementia. The quantification inaccuracy of a PET scanner can be attributed to a number of degrading factors and it is still unclear that which one plays the most important part and how to recovery the quantitative accuracy accordingly. The goal of this proposal is to greatly improve the quantitative accuracy of a PET scanner. We achieve the goal by using a model-based imaging reconstruction algorithm and the model is built from the following two-fold: (1) A correlated detector model is introduced to mitigate the indeterminate depth of interaction (DOI) of gamma rays within PET detector, and (2) accurate photon attenuation compensation is determined by using an attenuation map, which was provided by dual-energy material decomposition in a low-dose scenario. Different from the previous methods that use DOI detectors or system matrix modeled from either Monte Carlo simulation or point source measurement, the method in this proposal tries to decouple the degrading factors in the PET imaging processes, and models the factors which have the most severe impact on PET performance. This enables us to improve and quantitative accuracy of the PET scanner accordingly. To this end, model-based PET image reconstruction algorithms will be developed with the incorporation of photon attenuation map and the correlated detector model. The algorithm may provide a promising solution for quantification recovery and may represent a major step toward quantitative PET imaging.

定量精度的缺失是制约正电子发射断层成像仪(Positron Emission Tomography,简称PET)在肿瘤诊疗及其他重大疾病上应用的最重要的原因。普遍认为它与符合光子探测过程中的诸多物理因素有关。但是这些因素各自在多大程度上影响PET定量精度，且如何针对这些因素提出有效的方法恢复PET定量精度是学术界和工业界共同关心的课题。本项目从PET成像理论基础出发，分析制约PET定量精度的各种物理因素，围绕其中最主要的物理过程，即光子在物体中的衰减和在晶体条中作用深度不确定性展开研究。立足于对三维空间中符合光子与物体和晶体条相互作用的精确描述，构造与PET符合数据相匹配的物理模型，应用基于这些模型的重建算法，研究不同探测器结构和成像条件下PET的定量精度，并与不用模型的结果和真实的结果进行比对，评估算法的有效性，明确光子衰减、光子与晶体作用深度不确定性与PET图像定量精度之间的关系。

正电子发射断层成像仪（positron Emission Tomography, 简称PET）由于其分子成像和功能成像的特点，在肿瘤及其他重大疾病的诊疗上有着巨大的优势和应用前景，但由于符合光子探测中存在的各种物理因素导致了PET定量精度的缺失，使得PET重建图像无法准确反映活体真实的代谢分布，从而极大的制约了PET的发展。本项目针对PET图像定量精度恢复的问题，研究PET成像理论，分析制约PET定量精度中最主要的物理因素。项目组设计和开发了一套CT成像系统，提出了一种能应用于多能CT的统一材料分解架构和多能材料分解算法，减少了基材料分解中由于分解算法导致的伪影，提升了CT图像用于PET衰减校正的准确度，从而有效的恢复了PET的定量精度。该算法耗时短，不依赖于硬件条件，灵活性和通用性更强，显著提升了PET的定量精度，提高了PET系统的性能，扩展了PET仪器在重大疾病诊疗应用上的深度和广度。