基于PET/CT探测器灵敏度特性的呼吸运动伪影校正研究

Image degradation caused by patient's respiratory motion in PET/CT imaging may affect clinic diagnosis. The exiting methods for respiratory motion correction are mainly based in the space of reconstructed image to estimate motion parameters by image pixels, and then the estimated parameters can be used to correct motion by gated data, which still has some limitations. The proposed project with the novel method intends to utilize the property of detector geometric sensitivity relating to frame counts to estimate the parameters of respiratory motion, the parameters can be then applied to collected coincidences to correct motion. The idea of the proposed project has not yet been reported on any publication. Some interesting results in our pre-research experiment indicate that the shape of frame count cure is similar with the shape of respiratory motion cure. The evidence that the similarity between the two cures is increasing with the decreasing of interval time among the frames was observed in the pre-research. We suppose that the distribution of frame counts is closely associated with the respiratory parameters. To validate the suppose, the proposed is going through four steps: Monte Carol computation, physical phantom testing, animal model testing, and clinic testing, to deeply study the relationship between the frame counts and the respiratory parameters. Our research group has been working on the related project with having some published papers; we thus have had experience and the basis to study the proposed project. The project will explore a solution for respiratory motion correction on PET/CT imaging, it would be simple and quick in the implementation, which is very significant to improving the quality of cancer diagnosis at the early stage.

PET/CT扫描过程中，患者的呼吸运动成像伪影会影响临床诊断的质量。现有去伪影的方法主要基于图像空间像素估算运动参数，再根据已门控数据校正运动伪影，尚存局限性。本课题拟从新的角度，用探测器灵敏度相关的帧光子数估计呼吸运动参数，再应用于符合事件校正伪影的方法目前尚无报道。本课题预研实验发现，帧光子数序列分布曲线与呼吸运动曲线存在相似性；并观察到随着帧光子序列密度的增加，二者之间的相似性就越高。我们推测帧光子数分布与运动参数关系密切，可用于运动估计校正。为此，课题拟经过蒙特卡罗计算、物理体模测试、动物模型测试和临床数据验证四个环节，深入研究帧光子数分布和运动参数之间的关系。课题组有前期的工作研究积累并发表过相关论文，具备了开展此课题的经验和基础。本课题将为PET/CT成像胸腹部器官呼吸运动伪影校正，探寻一种解决问题的途径。它与传统方法相比，简单速度快,对提高肿瘤患者的早期诊断质量很有意义。

在世界范围内，肿瘤的发病率居高不下。PET/CT扫描成像设备对肿瘤患者早期的临床诊断起到了非常重要的作用。然而，患者的自然呼吸运动会造成胸腹部器官PET/CT成像的运动伪影导致图像质量下降，这对临床医生的诊断和治疗带来了干扰。因此，本项目的目的就是要研究探索一种有效的方法，补偿患者呼吸运动照成的PET/CT图像质量下降的问题。传统的方法主要基于图像空间像素开展工作，补偿运动造成的图像伪影，有一些局限性。本项目是探索一种新的数据驱动方法，它基于 PET/CT 探测器灵敏度相关的帧光子数分布特性，估计出PET/CT 扫描成像过程中呼吸运动的参数。另外，研究基于PET/CT 扫描采集 List-mode 数据模式中每时相帧内符合事件 LORs（Line of Response events）光子变化分布的规律，推断出与呼吸运动时相与配准参数之间的关系，对帧光子LORs 进行配准校正。我们基于探测器灵敏度，采用List-mode 数据模式，研究帧光子数分布特性，经过试验仿真和临床测试发现，帧光子数序列分布特性和与患者的呼吸运动呈现高度相关性，而且证明随着帧光子序列密度的增加，帧光子数分布与呼吸运动参数关系密切相关，并推断出基本的参数估计模型，可用于运动参数的估计，从而对每时相帧内符合事件 LORs（Line of Response events）光子进行配准校正。本项目研究的结果对PET/CT扫描呼吸运动图像质量的改进有积极探索促进作用。它的研究结果也可以应用到其它胸腹部内器官，例如肝和心脏等部位的运动伪影校正，经过一定的临床转化，对医院临床患者的诊断具有潜在的应用价值。