Data Consistency Based Motion Artifact Reduction for Head CT

描述(申请人提供)：大多数常规CT方法假定被成像的对象在扫描过程中是静止的。在重要的临床应用和生物医学研究中，如儿童头部CT，患者的运动往往是不可避免的，导致大量的运动伪影。该项目的长期目标是开发有效的基于运动估计的CT图像重建方法，提高诊断性能，并使新的临床应用成为可能。由于运动伪影的减少一直是CT中的一个主要问题，我们提出的工作可能会产生更好的图像质量，更准确地提取生理和病理特征，从而对医疗保健产生重大影响。 在这个R03项目中，我们将重点研究圆形扫描轨迹和面内头部运动，这不仅是我们感兴趣的情况下的一个令人满意的模型，也为我们未来研究更复杂的运动模式和扫描模式奠定了坚实的基础。具体目标是(1)建立运动模型，并从沿圆形轨迹收集的扇束/多切片数据估计相应的参数，以精确描述平面内运动；(2)采用广义锥束重建方法重建扇束/多切片几何结构中的运动对象；(3)在数值模拟、体模实验和回顾性患者研究中对所提出的方法进行评估。 完成后，所提出的技术将得到优化，并在数值模拟、体模实验和临床应用中展示其优越的性能。使用我们提出的方法，在有代表性的案例中，运动诱导的模糊将至少减少到与当前商业方法选择相关的30%。该项目将为后续的NIH R01提案生成试点数据，该提案以动态CT为目标，采用非环形扫描和一般主题运动。项目说明： 运动伪影的去除一直是CT中的一个主要问题。我们的项目将开发一种新颖而有效的解决方案来解决这个长期存在的问题，并可能产生明显更好的图像质量，以更准确和更稳健地提取生理和病理特征。因此，拟议的技术有望转化为临床使用，并应产生巨大的医疗保健效益，特别是对儿童。

DESCRIPTION (provided by applicant): Most of routine CT methods assume that a subject being imaged is stationary during the scan. In important clinical applications and biomedical researches, such as pediatric head CT, the patient-motion is often unavoidable, leading to significant motion artifacts. The long-term goal of this project is to develop effective CT methods for motion estimation based image reconstruction, improve the diagnostic performance, and enable new clinical applications. Because motion artifact reduction has been a major problem in CT, our proposed work may have a significant impact on the healthcare by producing better image quality and extracting physiological and pathological features more accurately. In this R03 project, we will focus on the circular scanning trajectory and in-plane head motion, which is not only a satisfactory model in the cases we are interested but also a solid basis for our future study on more complicated motion patterns and scan modes. The specific aims are to (1) develop motion models and estimate the corresponding parameters from fan-beam/multi-slice data collected along a circular locus to describe the in-plane motion precisely; (2) adapt generalized cone-beam reconstruction methods to reconstruct a moving object in the fan-beam/multi-slice geometry; and (3) evaluate the proposed methods in numerical simulations, phantom experiments and retrospective patient studies. On completion, the proposed techniques will have been optimized, and their superior performance demonstrated in numerical simulation, phantom experiments and clinical applications. Using our proposed methods, the motion-induced blurring will have been reduced down to at least 30% of that associated with the current commercial methods of choice in the representative cases. This project will generate pilot data for a follow-up NIH R01 proposal targeting dynamic CT with noncircular scanning and general subject motion.Project Narrative: Motion artifact reduction has been a major problem in CT. Our project will develop a novel and effective solution to this long-standing problem, and may produce significantly better image quality to extract physiological and pathological features more accurately and more robustly. Hence, the proposed techniques promise to be translated for clinical use and should generate great healthcare benefits, especially for children.