【作者】 刘晶晶；

【导师】 谢庆国；

【作者基本信息】 华中科技大学 ， 生物医学工程， 2015， 博士

【摘要】 正电子发射断层扫描(PET)是一种核医学成像设备,在肿瘤的早期检测、诊断等方面具有重大的应用价值。PET的普遍应用对成像质量提出了更高的要求。为了提高PET成像质量,获得高系统空间分辨率一直是PET仪器研制的重要方向之一。空间分辨率越高,意味着病灶检出率越高,越可能更早期发现微小病灶。传统PET仪器的研究和开发,主要集中在针对广泛应用的通用模式技术上,成像仪构建完成后各项性能指标几乎完全固定,不能在应用需求变化时进行快速调整或修改。我们注意到,不同的应用对各项性能的需求有所不同,并且在应用中往往对局部区域非常感兴趣,仅需要在感兴趣区域(ROI)获得高质量成像。以数字化技术为核心的全数字PET探测器,采用单事件、模块化的设计,能便利地实现变结构成像。基于全数字PET探测器的特点,本文提出了一种新的具有局部高空间分辨率的PET系统设计(定义为空间异质数字PET系统),研究了探测器的布局与ROI的位置、大小和空间分辨率需求的关系,分析了系统在模拟的肝脏病灶检测应用中的成像特性,最后探索了一种采用薄平板源的快速简易归一化校正方法,为未来实现系统奠定基础。主要研究内容和成果如下：空间异质数字PET系统,由多种具有不同固有空间分辨率的PET探测器构成,在不牺牲成像视野(FOV)的情况下,采用变结构的方式在ROI获得高空间分辨率、非ROI获得普通空间分辨率,从而兼顾应用的整个成像需要。本文首先提出了空间异质数字PET系统设计思想和成像方法,然后设计了一种系统原型,包含高、低两种不同固有空间分辨率的探测器,具有高固有空间分辨率的探测器(HRD)在探测环上采用“一段式分布”,细致研究了该分布方式下HRD的布局(位置和数目)对FOV中各局部区域的空间分辨率的影响。同时设计了一种“均匀分布”的方式,与“一段式分布”形成互补,实现了对FOV中任意ROI的局部高空间分辨率成像。该项工作构建了ROI的空间分辨率需求、位置和大小与HRD布局的关系表,为ROI高空间分辨率成像实现了定制化的空间异质数字PET系统。以肝脏病灶检测为应用背景,采用蒙特卡罗(MC)仿真的方法,分析了“一段式分布”的空间异质数字PET系统的成像特性。具体研究了成像收敛特性,分析成像性能和随机事件对成像质量的影响。实验结果表明,HRD的布局直接影响ROI的成像收敛速度和成像性能,其影响规律与HRD布局对ROI空间分辨率影响的结论一致；并且系统相对于同质系统的成像性能基本不受随机分数高低的影响。为满足离线和在线变结构的空间异质数字PET系统对归一化校正速度的高要求,设计了一种基于薄平板源的快速简易归一化校正方法,对离线变结构系统须进行一次完整的归一化校正而对在线变结构系统仅须更新部分校正因子。经分析该方法同样适用于同质系统,故采用Trans-PET(?)系统进行了方法的初步验证。实验结果表明该方法能基本满足离线变结构的空间异质数字PET系统归一化校正的需要,结合对该方法的分析表明,该方法也有潜力为在线变结构的空间异质数字PET系统实现快速归一化校正。更多还原

【Abstract】 Positron emission tomography (PET) is a nuclear medicine instrument, which has shown great value in the early detection and diagnosis of tumor. As the wide application, PET is expected to provide better imaging quality. To improve the imaging quality of PET, acquiring high spatial resolution is one of the key directions in the development of PET instrument. Higher spatial resolution always means better lesion detectability to detect tiny lesions earlier.The research and development of traditional PET focus on the technology of common design mode for general purpose. After the implementation of PET instrument, the performances are almost fixed and can’t be adjusted or modified to fit the variation of the application’s requirements. We noticed that different applications always have different requirements and the focus point is always the region of interest (ROI). Usually, only the ROI requires high imaging quality. The all-digital PET detector adopts the design of single event and modularization. Digital technology is the core of the detector. The detector can be utilized to easily realize PET imaging with variable structures. Therefore, based on the characteristics of all-digital PET detectors, we proposed a new local high spatial resolution PET system in this thesis. We investigated the relation between the layout of the detectors and the spatial resolution requirement of ROI with different positions and sizes. Then the imaging characteristics of the application of the liver lesion detection were analyzed. At last, we initially investigated a new fast and simple normalization correction method based on thin planar source, which aimed at the real implementation of the proposed system. The overall summaries are outlined as following:The proposed system was composed of mixed detectors having different intrinsic spatial resolutions. To meet the requirement of application, the proposed system can provide high spatial resolution in the ROI and normal spatial resolution in the region outside of the ROI, by changing the layout of the detectors and without sacrificing the field of view (FOV). Firstly, the design concept and imaging method of the new PET was proposed. Then, we designed one kind prototype of the proposed system with two types of detectors, including normal resolution detectors (NRDs) and high resolution detectors (HRDs). The NRDs and HRDs were located on the detector ring, with all the HRDs (NRDs) distributed as one arc. The effects of the location and the size of the HRD arc on the spatial resolution of every sub-region in the FOV were analyzed. And we proposed an uniform distribution of the HRDs, which was a complementation to the one arc distribution. These two distributions led to local high spatial resolution for any ROI in the FOV. These work established the relation between the spatial resolution requirement of ROI with different position and size and the layout of the HRDs. Customized PET system with mixed detectors can be designed for high spatial resolution of ROI.Based on Monte Carlo (MC) simulation, we analyzed the imaging characteristics of the proposed system with one arc distribution in the application of liver lesion detection. The convergence properties, the imaging performances and the effect of the random events on the imaging quality were analyzed. The results shown that the convergence rate and the imaging performances of ROI directly related to the HRD layout, which was consistent with the conclusion of the effects of the HRD layout on the ROI’s spatial resolution. And the random fraction had little effect on the relative imaging performances of the proposed system with respect to the traditional system.To meet the high time requirement of normalization correction due to the off-line and on-line changement of system structure, we designed a fast normalization method based on a thin planar source. For the off-line, a single complete correction is needed to obtain all correction factors, and for the on-line, only part factors are needed to update. Because the method is also fit to the system composed with homogeneous detectors, the constructed Trans-PET(?) system was adopted to verify the method. The results has shown that this method can meet the basic requirement of short-time normalization for the off-line structure changement of the proposed system. Combination with the analysis of this method, it indicated that the method has the potential to provide fast normalization correction for the proposed system with on-line structure changement.更多还原