FAST FULLY 3D IMAGE RECONSTRUCTION USING PLANOGRAMS

使用平面图快速重建全 3D 图像

• 批准号: 6174190
• 负责人: Paul E. Kinahan
• 金额: $ 14.7万
• 依托单位: UNIVERSITY OF PITTSBURGH AT PITTSBURGH
• 依托单位国家: 美国
• 财政年份: 1999
• 资助国家: 美国
• 起止时间: 1999-04-01 至 2002-03-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6174190
• 关键词: bioimaging /biomedical imaging; diagnosis design /evaluation; image enhancement; mathematical model; model design /development; neoplasm /cancer radionuclide diagnosis; positron emission tomography; single photon emission computed tomography; technology /technique development

DESCRIPTION (Adapted from Applicant's Abstract): The goal of this proposal is to implement and comparatively evaluate a reconstruction algorithm based on a new data acquisition format called a "planogram" that is applicable to PET scanners with flat detector surfaces. The planogram algorithm uses a data acquisition format that is intrinsic to dual-head SPECT scanners operated in coincidence mode. These scanners have received considerable attention as cost-effective alternatives to PET scanners for whole-body oncology and are in use at an ever increasing number of hospitals. There is thus a large base of existing scanners that can take advantage of fast and accurate reconstruction algorithm. Current reconstruction methods are unable to generate images from coincident-mode SPECT scanner data in a feasible time without trade-offs in accuracy or propagation of statistical noise. The applicants described having derived mathematical relationships for the planogram format that, while conceptually complex, are computationally simple and allow the reconstruction computations to be performed entirely with fast Fourier transforms (FFTs). This will allow coincidence-mode SPECT systems, and PET scanners with a flat similar plate detector geometry, to realize their full sensitivity while still reconstructing images in a clinically feasible time. It is expected that the planogram algorithm can be optimized to reduce image reconstruction time by about two orders of magnitude, compared to current techniques, without any increase in noise propagation or loss of resolution. The specific aims of this proposal are to (i) implement and validate the planogram reconstruction algorithm and (ii) compare the planogram algorithm to standard 3D reconstruction methods in terms of reconstruction time, accuracy, and reduction o statistical noise. The long-term objective of this research is to assist in the development of cost-effective whole-body PET scanners for clinical oncology imaging, for which the image reconstruction is a critical component.

描述（改编自申请人的摘要）：本项目的目标 建议是实施并比较评估重建 基于称为“货架图”的新数据采集格式的算法 适用于具有平坦探测器表面的 PET 扫描仪。这 货架图算法使用固有的数据采集格式 双头 SPECT 扫描仪在重合模式下运行。这些扫描仪 作为具有成本效益的替代品而受到广泛关注 用于全身肿瘤学的 PET 扫描仪的使用率非常高 医院数量不断增加。因此现有的基础设施很大 可以利用快速准确重建的扫描仪 算法。目前的重建方法无法生成图像 在可行的时间内从重合模式 SPECT 扫描仪数据中获取数据，无需 准确性或统计噪声传播的权衡。这 申请人描述了已经导出的数学关系 货架图格式虽然概念上很复杂，但计算起来却很复杂 简单并允许执行重建计算 完全通过快速傅立叶变换 (FFT) 进行。这将允许 符合模式 SPECT 系统和 PET 扫描仪具有类似的平面 板探测器几何形状，以实现其全部灵敏度，同时仍然 在临床可行的时间内重建图像。预计 可以优化货架图算法以减少图像重建 与现有技术相比，时间缩短了大约两个数量级， 不会增加任何噪声传播或分辨率损失。这 该提案的具体目标是 (i) 实施并验证 货架图重建算法和 (ii) 比较货架图 标准 3D 重建方法的算法 重建时间、准确性和统计噪声的减少。这 这项研究的长期目标是协助开发 用于临床肿瘤成像的经济高效的全身 PET 扫描仪， 其中图像重建是一个关键组成部分。

项目成果

Fast fully 3-D image reconstruction in PET using planograms.

使用货架图在 PET 中快速进行全 3D 图像重建。

• DOI: 10.1109/tmi.2004.824231
• 发表时间: 2004
• 期刊: IEEE transactions on medical imaging.
• 作者: Brasse,D;Kinahan,PE;Clackdoyle,R;Defrise,M;Comtat,C;Townsend,DW
• 通讯作者: Townsend,DW