Iterative Image reconstruction for high-resolution PET imaging

描述（由申请人提供）：已开发出用于发射断层摄影的迭代重建算法，其相对于滤波反投影方法显著提高了图像质量。然而，大多数当前的重建算法隐含地假设系统模型是精确的。在每个前向和后向投影中直接使用精确的系统模型所带来的令人生畏的计算挑战往往导致人们采用不太精确的模型。这导致重建图像中的噪声增加和分辨率降低，因为在现有方法中不能校正建模误差的影响。这项资助的目标是开发一类新的迭代重建方法，可以补偿建模误差的影响。这项工作是基于我们的深入分析，从系统模型中的每个组件到重建图像的误差传播。新方法的创新之处在于，它不需要在每一个前向和后向投影精确的系统模型。当在迭代重建中直接使用精确的系统模型不切实际时，该方法可以获得高分辨率的图像，并且它还可以通过使用简化的快速正向投影仪和反向投影仪来减少重建时间而不牺牲图像质量。我们将首先发展具有误差校正能力的高分辨率迭代图像重建的理论。然后，我们将集中在应用和验证的理论在正电子发射断层扫描（PET）。我们将在microPET扫描仪上实施新的重建算法，并将使用Monte Carlo模拟、物理体模实验和真实的动物数据来评估病变检测和定量性能。我们相信，新算法将提供高分辨率的图像和准确的定量信息，以了解小动物模型中的人类疾病。一旦成功，我们将重建算法扩展到临床成像系统，也将应用该理论的其他成像方式，如X射线CT，SPECT，MRI和光学断层扫描。摘要：正电子发射断层扫描（PET）是一种功能成像方式，广泛用于临床和生物学研究。本计画将开发一种新的PET影像重建方法，提供高解析度的影像及精确的量化资讯，以了解及治疗人类疾病。

DESCRIPTION (provided by applicant): Iterative reconstruction algorithms that significantly improve image quality over filtered backprojection methods have been developed for emission tomography. However, most current reconstruction algorithms implicitly assume that the system model is exact. The daunting computational challenge associated with the direct use of an exact system model in each forward and back projection has often led people to adopt less accurate models. This results in increased noise and reduced resolution in reconstructed images, because the effect of the modeling error cannot be corrected in the existing methods. The goal of this grant is to develop a new class of iterative reconstruction methods that can compensate the effect of modeling error. The work is based on our thorough analysis of error propagation from each component in the system model into reconstructed images. The innovation of the new method is that it does not require an exact system model in every forward and back projection. The method can obtain high-resolution images when direct use of an accurate system model in the iterative reconstruction is impractical, and it can also reduce reconstruction time by using simplified fast forward and back projectors without sacrificing image quality. We will first develop the theory of high-resolution iterative image reconstruction with error correction capability. Then we will focus on the application and validation of the theory in positron emission tomography (PET). We will implement new reconstruction algorithms on microPET scanners, and will evaluate the lesion detection and quantitation performance using Monte Carlo simulations, physical phantom experiments, and real animal data. We believe that the new algorithms will provide high-resolution images and accurate quantitative information for understanding human diseases in small animal models. Upon success, we will extend the reconstruction algorithm to clinical imaging systems and will also apply the theory to other imaging modalities, such as X-ray CT, SPECT, MRI, and optical tomography. Lay abstract: Positron emission tomography (PET) is a functional imaging modality that is widely used in clinical and biological studies. This project will develop a novel image reconstruction method for PET which will provide high-resolution images and accurate quantitative information for understanding and treating human diseases.