High performance PET Detector Module for Human Brain Imaging

项目总结/摘要 本提案是对PA 18-484的回应。在BRAIN 2025报告中，PET（正电子发射断层扫描）是 被认为是“将神经递质、受体和神经调质的研究转化为 人类”然而，受体占有率和代谢的动态评估受到空间分布的阻碍。 分辨率和灵敏度，甚至是最现代的临床可用的PET扫描仪。为了解决这个 挑战，我们提出了一种具有高度创新设计的下一代PET探测器： 分层闪烁体结构和侧面读出配置。模块中的水晶板堆叠在一起 沿着深度方向，并通过反射膜光学分离。在每一个创造的闪烁光 通过位于晶体的四个侧面上的光电探测器来测量层的厚度。与传统PET相比 探测器，其中包含像素化的晶体阵列，新的设计具有以下优点：（1）分层 一种结构提供相互作用深度（DOI）信息，使得可以使用较小直径的检测器环 而不增加视差，在降低成本的同时增加灵敏度。(2)四边读出法 通过降低闪烁光收集效率来提高系统的能量分辨率 由于全内反射造成的光损失。(3)亚毫米空间分辨率是可以实现的，而无需使用非常 小间距晶体阵列，因为每个晶体层中的相互作用位置是通过机器学习确定的， 基于在四个晶体侧面上收集的光分布的解码。因此， 晶体会减少。(4)由于每个层的相互作用位置和能量分辨率是确定的， 独立地，可以通过在模块中堆叠更多层来增加系统灵敏度，而不影响 系统的空间和能量分辨率。(5)对于侧面读出设置，较大的横截面积比 为了覆盖模块的所有四个侧面，需要较少的光电探测器;这降低了光电探测器的成本。 上述前四点已在使用小型原型模块的初步研究中得到证明。我们 建议用这种新设计建造一个大型探测器模块来验证第五个优势，并进行研究 探测器尺寸对前四个的影响。该提案的结果将是两个DOI启用探测器模块 具有出色的空间分辨率（~1 mm）和能量分辨率（~10%），以及良好的时序分辨率（~400 ps）、DOI分辨率（~ 3 mm）和高系统灵敏度。两个模块的完整表征研究 对德伦佐和霍夫曼大脑幻影的成像研究将解决目前的局限性， 人脑PET扫描仪，并将作为一个新的动态PET扫描仪的神经成像的基础。

Project Summary / Abstract This proposal is in response to PA 18-484. In the BRAIN 2025 Report, PET (positron emission tomography) is identified as “the best means to translate studies of neurotransmitters, receptors, and neuromodulators to humans.” However dynamic assessment of receptor occupancy and metabolism is hindered by the spatial resolution and sensitivity of even the most modern of clinically available PET scanners. To address this challenge, we propose a next generation PET detector with a highly innovative design: a detector module with a layered scintillator structure and a side readout configuration. The crystal slabs in the module are stacked along the depth direction and are optically separated by reflective films. The scintillation light created in each layer is measured by photodetectors located on the four sides of the crystal. Compared with traditional PET detectors, which contain pixelated crystal arrays, the new design has the following advantages: (1) The layered structure provides depth of interaction (DOI) information such that a smaller diameter detector ring can be used without increasing parallax error, increasing sensitivity while lowering costs. (2) The four-sided readout method improves the energy resolution of the system with increased scintillation light collection efficiency by reducing light loss due to total internal reflection. (3) Sub millimeter spatial resolution is achievable without using very small pitch crystal arrays, since the interaction location in each crystal layer is determined via machine learning- based decoding of the light distribution collected on the four crystal sides. Therefore the production cost of the crystals is reduced. (4) Since the interaction location and energy resolution for each layer are determined independently, the system sensitivity can be increased by stacking more layers in the module without affecting the spatial and energy resolution of the system. (5) For side readout setup, a larger ratio of cross-sectional area to length requires fewer photodetectors to cover all four sides of the module; this reduces photodetector cost. The first four points above have been demonstrated in preliminary studies using a small, prototype module. We propose to build a large scale detector module with this new design to verify the fifth advantage, and to study the effect of detector size on the first four. The outcome of this proposal will be two DOI enabled detector modules with excellent spatial resolution (~1 mm) and energy resolution (~10%), as well as good timing resolution (~400 ps), and DOI resolution (~ 3 mm) and high system sensitivity. A full characterization study for the two modules and imaging studies for both the Derenzo and Hoffman brain phantoms will address the current limitations of human brain PET scanners, and will serve as the foundation for a new dynamic PET scanner for neuroimaging.