Novel algorithm for improved contrast enhanced cardiac MRI

描述(由申请人提供)：心肌首过灌注和晚期Gd增强(LGE)计划是大多数临床心脏MRI检查的关键组成部分。现有磁共振成像方案的局限性往往使其难以同时获得高时空分辨率、足够的空间覆盖率和良好的图像质量，这使得对结果的解释变得困难。同样，大量的屏气和长时间的屏气往往使LGE采集对许多患者具有挑战性，导致显著的运动伪影和患者吞吐量减少。在这种背景下，临床上迫切需要一种新的动态成像框架，能够在不降低质量的情况下实现自由呼吸采集并显著提高时空分辨率和覆盖范围。该建议的主要目标是开发一种新的动态成像框架，该框架可以实现自由呼吸的心脏MRI，并以最小的伪影显著地加速它。我们最近引入了一种新的正则化重建算法来显著提高自由呼吸动态MRI数据的速度。算法的初步验证表明，所提出的方案能够提供高达11倍的速度和较小的伪影。本方案的主要重点是进一步完善k-t SLR方案，实现临床高分辨率心肌灌注和自由呼吸LGE MRI。拟议研究的成功完成将提供时间分辨率为一次心跳、整个心脏的空间分辨率为0.15x0.15x0.8cc的定量血流灌注估计，这比目前的方案提高了四倍。同样，我们希望通过放宽屏气要求和减少LGE MRI数据的扫描时间来显著改善患者的依从性。这些进展意义重大，将极大地提高对比增强CMRI的技术水平。该算法与依赖于x-f空间稀疏性的经典方法有很大的不同。此外，我们还引入了非凸谱先验知识，并利用动态图像的稀疏性进一步提高了数据的保真度和加速比。因此，拟议的方案具有很高的创新性，其影响预计将延伸到具体应用之外。我们的团队完全有资格执行拟议的研究，因为我们在专业知识(包括信号/图像处理、磁共振物理学、放射学和心脏病学)方面的综合范围和广度，以及广泛的初步数据。 公共卫生相关性：拟议的项目致力于开发一种新的采集和数据处理方案，以提高对比增强心脏核磁共振的性能。这项研究与公共卫生相关，因为该方案可以显着改善数据的解释，并提高患者的依从性和舒适性。此外，扫描时间的减少将提高吞吐量。因此，这些发现最终有望应用于改善人类健康。

DESCRIPTION (provided by applicant): Myocardial first-pass perfusion and late gadolinium enhancement (LGE) schemes are key components of most clinical cardiac MRI exams. The limitations of current MRI schemes often makes it challenging to simultaneously achieve high spatio-temporal resolution, sufficient spatial coverage, and good image quality in first-pass perfusion MRI, making it difficult to interpreting the results. Similarly, the large number of breath-holds and their long duration often makes LGE acquisitions challenging for many patients, resulting in significant motion artifacts and reduced patient throughput. In this context, there is an immediate clinical need for a novel dynamic imaging framework that can enable free-breathing acquisitions and considerably improve spatio-temporal resolution and coverage, without degrading the quality. The main objective of this proposal is to develop a novel dynamic imaging framework, which can enable free-breathing cardiac MRI and significantly accelerate it with minimal artifacts. We recently introduced a novel regularized reconstruction algorithm to significantly accelerate free-breathing dynamic MRI data. Preliminary validations of the algorithm demonstrated the ability of the proposed scheme to provide accelerations of up-to eleven fold with minor artifacts. The main focus of this proposal is to further improve the k-t SLR scheme and use it to realize high-resolution clinical myocardial perfusion and free-breathing LGE MRI. The successful completion of the proposed research will provide quantitative perfusion estimates with a temporal resolution of one heartbeat and spatial resolution of 0.15x0.15x0.8 cc from the entire heart, which is a four-fold improvement over current schemes. Similarly, we expect to considerably improve the patient compliance by relaxing the breath-holding requirement and reducing the scan time in LGE MRI data. These developments are quite significant and will considerably advance the state of the art in contrast-enhanced CMRI. The proposed algorithm is a radical departure from the classical approaches that rely on x-f space sparsity. In addition, we introduce non-convex spectral priors and additionally exploit the sparsity of the dynamic images to further improve the data fidelity and acceleration rate. Thus, the proposed scheme is highly innovative and its impact is expected to extend beyond the specific applications. Our team is well qualified to perform the proposed research because of our combined scope and breadth in expertise (including signal/image processing, MR physics, radiology, and cardiology), in addition to the extensive preliminary data. PUBLIC HEALTH RELEVANCE: The proposed project addresses the development of a novel acquisition and data-processing scheme to improve the performance of contrast enhance cardiac MRI. This research has relevance to public health since this scheme can significantly improve the interpretation of the data and improve patient compliance and comfort. In addition, a reduction in scan time will improve throughput. Thus, the findings are ultimately expected to be applicable to improve the health of human beings.