Supplemental transmission aided attenuation correction for high performance quantitative PET imaging

项目总结 来自PET-CT和PET-MR组合的定量示踪图像正越来越多地被用于放射学 决策和临床试验。基于CT和MR的PET光子衰减数据校正，通常是 对示踪剂定量的最大影响，可导致受试者辐射剂量增加或需要较长时间 分别进行额外的专用磁共振扫描。使用PET信号的人工智能算法 仅从受试者身上来估计衰减已经显示出有希望的早期结果，但是，可以有 当应用于与训练数据不同的考试时，性能降低。重建方法 联合估计从对象PET信号测量的衰减和示踪剂对比度代表 可行的替代方法，但到目前为止，未能与CT方法的量化相匹配。因此，不 用于PET衰减校正的单一方法，始终产生高量化而不是 危及患者安全或目前存在的吞吐量。这是一个重要的问题，因为研究和 临床结果必须在这些实际问题和数据质量之间取得平衡。这项提议的总体目标是 开发和表征一种高性能的衰减校正方案，该方案利用PET 来自受试者的信号和可变的活动外部来源。中心假设是，拟议的 与使用PET的重建算法相比，重建方法的性能将有显著的提高 单独来自对象的信号，与成像研究无关。两个具体目标包括：1) 一种基于补充传输的PET图像光子校正算法的开发与验证 商业PET成像过程中的衰减和2)原型和表征能够 动态变化的外部来源活动。目标1结果将是一个重建算法，即 经过优化，可为一些最常见的临床PET研究提供定量的PET数据。在目标2下，a 可重复性的原型设备可以改变外部源活动，以最大限度地利用示踪剂 量化，同时将不可避免的患者示踪图像噪声降级降至最低 任何外部来源的引进，都会产生。该创新是一种衰减校正策略， 极大地缓解了以前的关节重建方法的局限性，提供定量的PET诊断 预计将与银色标准CT方法相匹配。这一点意义重大，因为 接受脑部和心脏PET-CT扫描的患者，检查建议的方法预计将受益， 稳步增长。对于PET-MR，该方法可以在MR有限的地方改进示踪剂定量 通过消除仅使用非诊断性衰减的MR来提高性能并增加患者吞吐量 收购。因此，拟议的战略具有显著改善放射决策的巨大潜力- 根据定量PET摄取测量得出的临床试验结果。

PROJECT SUMMARY Quantitative tracer images from combined PET-CT and PET-MR are increasingly being utilized for radiological decision-making and clinical trials. CT and MR based data corrections for PET photon attenuation, typically the largest impact on tracer quantification, can lead to increased radiation dose to the subject or require lengthy additional dedicated MR scans, respectively. Artificial intelligence algorithms that use the PET signal originating from the subject alone to estimate attenuation have shown promising early results, but, can have reduced performance when applied to exams different from the training data. Reconstruction methods that jointly estimate both the measured attenuation and tracer contrast from the subject PET signal represent a viable alternative, but, to date, have failed to match the quantification of CT approaches. Consequently, no single approach for PET attenuation correction that consistently produces high quantification and does not compromise patient safety or throughput currently exists. This is an important problem, since research and clinical findings must balance these practical issues with data quality. The overall objective of this proposal is to develop and characterize a high performance attenuation correction scheme that utilizes both the PET signal from the subject and a variable activity external source. The central hypothesis is that the proposed reconstruction method will have significantly improved performance over reconstruction algorithms using PET signal originating from the subject alone, independent of the imaging study. The two specific aims include: 1) developing and validating a supplemental transmission-based algorithm for correcting PET images for photon attenuation during commercial PET imaging and 2) prototyping and characterizing a device capable of dynamically varying external source activity. The result of Aim 1 will be a reconstruction algorithm that is optimized to produce quantitative PET data for some of the most common clinical PET studies. Under Aim 2, a prototype device that can repeatability vary the external source activity in order to maximize tracer quantification while minimizing the unavoidable degradation to patient tracer image noise, caused by the introduction of any external source, will be produced. The innovation is an attenuation correction strategy that greatly mitigates the limitations of previous joint reconstruction methods to deliver quantitative PET diagnostics that are expected to match those of silver standard CT approaches. This is significant because the number of patients receiving brain and cardiac PET-CT scans, exams the proposed method is expected to benefit, is steadily increasing. For PET-MR, the method may improve tracer quantification where MR has limited performance and increase patient throughput by eliminating the need for non-diagnostic attenuation-only MR acquisitions. Thus, the proposed strategy has great potential to significantly improve radiological decision- making and clinical trial findings that rely on quantitative PET uptake measurements.