Model-Based Reconstruction for Dynamic MRI

描述（由申请人提供）：生物计算和MRI采集技术的进步使得可以考虑将比标准傅立叶逆变换更复杂的图像重建方法应用于MRI数据。这将对动态MRI产生深远的影响。提出了将时间和空间模型纳入动态对比增强（DCE）MRI数据的逆问题框架内的重建。具体目标是：（1）在稀疏采样动态MRI数据集的重建方法中开发并纳入低级别（对强度随时间变化的约束）和高级别（参数化）时间模型。这些模型将允许体积覆盖的大幅增加，而不会伴随SNR降低。(2)为稀疏采样的动态MRI数据集开发基于空间模型的重建方法。低层空间模型将实现空间约束，更高层的空间模型将从患者特定的空间参考数据创建。(3)扩展基于模型的时空采集和重建方法，以适应患者运动。(4)将基于模型的方法扩展为联合收割机与多线圈加速（并行成像）方法相结合。(5)为心肌灌注磁共振成像的临床应用提供建议的计算方法，并为更广泛的研究社区提供数据和软件工具。研究方法：我们的多学科团队将开发软件工具作为一个协作过程，将生物计算和MRI专业知识与临床心脏成像专业知识相结合。将使用基于模型的多线圈方法重建心脏灌注数据的笛卡尔和径向简化k空间采集并进行比较。将使用软件方法或呼吸带和预扫描校准识别和补偿呼吸运动。由此产生的软件工具将纳入ITK，供研究界使用。与公共卫生相关的是，心脏病是死亡的主要原因。这一建议提供了新的重建方法，将推进动态MRI领域，提高心肌血流的无创评估。这些改进将使心脏病得到更好、更及时的治疗和监测。所提出的方法可以扩展到改善非心脏的动态MRI应用，如肿瘤治疗的反应和大脑刺激的反应的研究。

DESCRIPTION (provided by applicant): Advancements in biocomputing and MRI acquisition technologies make it possible to consider applying image reconstruction methods that are more complex than the standard inverse Fourier transforms to MRI data. This will have a profound impact on dynamic MRI. Temporal and spatial models are proposed to be incorporated into the reconstruction of dynamic contrast enhanced (DCE) MRI data within an inverse problem framework. Specific aims are (1) Develop and incorporate low level (constraints on changes of intensity over time) and higher level (parameterized) temporal models within reconstruction methods for sparsely sampled dynamic MRI datasets. These models will allow for a large increase in volume coverage without concomitant SNR reductions. (2) Develop spatial model-based reconstruction methods for sparsely sampled dynamic MRI datasets. Low level spatial models will realize spatial constraints, and higher level spatial models will be created from patient-specific spatial reference data. (3) Extend the model-based spatio-temporal acquisition and reconstruction methods to accommodate patient motion. (4) Extend the model-based methods to combine with multi-coil speedup (parallel imaging) methods. (5) Validate the proposed computational methods for the clinical application of myocardial perfusion MR imaging and provide data and software tools to the broader research community. Methods: Our multi-disciplinary team will develop software tools as a collaborative process combining biocomputing and MRI expertise with clinical cardiac imaging expertise. Both Cartesian and radial reduced k-space acquisitions of cardiac perfusion data will be reconstructed with the model-based multi-coil methods and compared. Respiratory motion will be identified and compensated using either software approaches or a respiratory strap and pre-scan calibrations. The resulting software tools will be integrated into ITK and provided for use to the research community. The relevance to public health is that heart disease is the leading cause of death. This proposal offers new reconstruction methods that will advance the field of dynamic MRI and improve the non-invasive assessment of myocardial blood flow. Such improvements will allow better and more timely treatments and monitoring of heart disease. The proposed approach can be extended to improve non-cardiac dynamic MRI applications such as studies of the response of tumors to therapy and the response of the brain to stimuli.