Rapid motion-robust quantitative DCE-MRI for the assessment of gynecologic cancers

项目摘要 妇科癌症是影响妇女的一些最致命的疾病。全球范围内，一名妇女死于 每两分钟就得一次宫颈癌。MRI越来越多地用于妇科和许多其他疾病的评估。 癌的除了其用于癌症分期的既定用途之外，长期以来一直对使用MRI衍生的肿瘤标志物感兴趣。 定量指标，以深入了解肿瘤微环境。参数映射从 动态对比增强（DCE）MRI数据的定量可用于研究肿瘤血管分布并识别肿瘤血管。 灌注和氧合更好的肿瘤，因此对某些治疗更敏感， 化疗和放疗。然而，当前MRI的相对慢的成像速度和运动灵敏度 技术导致DCE-MRI数据的不可靠和不可再现的量化，这限制了其 在临床实践中的应用。 我们的团队是快速抗运动DCE-MRI技术开发的世界领导者，特别是使用 径向成像和压缩传感的组合。我们开发了一种叫做GRASP的技术， 被认为是一个学术和工业的伙伴关系，现在已经成功地转化为标准 临床实践虽然功能强大，但第一代GRASP也有局限性。首先，径向成像是鲁棒的 运动，但不是自由的运动，这通常会导致模糊。其次，GRASP使用非常简单的稀疏化 压缩感知的变换，这可能会引入量化问题。第三，巨猿生存方案不是 最初是为药代动力学分析而开发的，错过了重要的成分，如AIF的整合 估计和T1映射。第四，图像重建时间仍然很长-大约几分钟。 我们已经开发了新的进展来规避这些限制，并提供了一种新的DCE-MRI技术， 增加速度、运动阻力和个性化AIF估计以及药代动力学T1标测 分析.根据PAR-18-009指导方针，我们的主要目标是建立学术-工业伙伴关系 在纪念斯隆凯特琳癌症中心和通用电气医疗保健之间， 用于妇科和其他类型癌症患者的定量DCE-MRI的发展。具体 目标如下： 1.开发和实施超越GRASP的快速抗运动定量DCE-MRI技术 提供更高的速度和运动阻力;动态T1标测;以及个性化和自动化 药代动力学分析 2.评价快速运动的重复性、再现性和初步肿瘤反应评估- 稳健的定量DCE-MRI技术（“DCE-new”），并将DCE-new与护理标准DCE-MRI进行比较 （“DCE标准”） 3.开发和评估基于深度学习的快速图像重建算法

PROJECT SUMMARY Gynecologic cancers are some of the most lethal diseases affecting women. Globally, one woman dies of cervical cancer every two minutes. MRI is increasingly used in the evaluation of gynecologic and many other cancers. Beyond its established use for cancer staging, there has long been an interest in the use of MRI-derived quantitative metrics to gain insights into the tumor microenvironment. Parametric maps obtained from quantification of dynamic contrast enhanced (DCE) MRI data can be used to study tumor vascularity and identify tumors that are better perfused and oxygenated and thus more sensitive to some treatments such as chemotherapy and radiation. However, the relative slow imaging speed and motion sensitivity of current MRI technology results in non-reliable and non-reproducible quantification of DCE-MRI data, which restricts its application in clinical practice. Our group is a world leader in development of rapid motion-resistant DCE-MRI techniques, in particular using combinations of radial imaging and compressed sensing. We developed the technique called GRASP, which was conceived as an academic-industrial partnership and has now been successfully translated into standard clinical practice. Though powerful, the first generation of GRASP has limitations. First, radial imaging is robust to motion, but not free of motion, which usually results in blurring. Second, GRASP uses a very simple sparsifying transform for compressed sensing, which can introduce issues with quantification. Third, GRASP was not originally developed for pharmacokinetic analysis and misses important ingredients such as integration of AIF estimation and T1 mapping. Fourth, image reconstruction time is still very long – in the order of several minutes. We have developed new advances to circumvent these limitations and offer a new DCE-MRI technique with increased speed, motion-resistance and personalized AIF estimation and T1 mapping for pharmacokinetic analysis. Following the PAR-18-009 guidelines, our main goal is to form an academic-industrial partnership between Memorial Sloan Kettering Cancer Center and General Electric Healthcare to translate these new developments in quantitative DCE-MRI for use in patients with gynecologic and other type of cancers. Specific Aims are as follows: 1. Develop and implement a fast motion-resistant quantitative DCE-MRI technique that goes beyond GRASP to offer increased speed and resistance to motion; dynamic T1 mapping; and personalized and automated pharmacokinetic analysis 2. Evaluate the repeatability, reproducibility and preliminary tumor response assessment of the fast motion- robust quantitative DCE-MRI technique (“DCE-new”) and compare DCE-new to standard of care DCE-MRI (“DCE-standard”) in patients with gynecologic cancer 3. Develop and evaluate fast image reconstruction algorithms based on deep learning