Improved cardiac and vascular MRI using parallel imaging and compressed sensing

描述（由申请人提供）：本提案的目的是通过将非笛卡尔并行成像技术的新概念与新出现的压缩采样理论相结合，在心脏和血管成像的成像速度和SNR方面产生新的增益水平。压缩感知有望通过打破成像时间和SNR之间的传统联系来彻底改变MRI领域。在这里，我们将利用这些概念来开发一套全新的成像策略，大幅提高SNR和成像速度。我们通过为高端图形处理单元开发一个开源软件发行版来专门解决计算限制问题。这些处理器有望大幅减少医学成像中的计算时间。最终，我们相信，这些技术，当被视为一个整体，将导致一类新的方法，用于心脏和血管诊断，这将提供提高图像质量，信噪比和速度在MRI，也许是无与伦比的MRI的发展，导致显着改善成像的磁共振血管造影，心脏功能和心脏灌注。我们的具体目标是：1）开发和评估用于获取和重建用于2D MRI应用的多层非笛卡尔并行成像方法的改进方法2）开发和评估鲁棒的组合非笛卡尔并行成像和压缩感测方法3）开发和评估用于计算非笛卡尔并行成像的基于图形处理单元（GPU）的改进计算方法，CG-HYPR和组合方法，以实现临床可接受的重建时间; 4）验证并行CG-HYPR方法用于评价心血管疾病，以缩短总检查时间并提高图像质量。

DESCRIPTION (provided by applicant): The objective of this proposal is to produce a new level of gains in imaging speed and SNR for cardiac and vascular imaging by combining novel concepts of non-Cartesian parallel imaging techniques with the newly emerging compressed sampling theory. Compressed sensing promises to revolutionize the field of MRI by breaking the traditional link between imaging time and SNR. Here we will exploit these concepts to develop a set of completely new imaging strategies with dramatic increases in SNR and imaging speed. We specifically address computational limitations by developing an open source software distribution for high-end graphical processing units. These processors promise to dramatically reduce computational time across the board in medical imaging. Ultimately we believe that these technologies, when viewed as a whole, will result in a novel class of methods for cardiac and vascular diagnosis which will provide an increase in image quality, SNR and speed in MRI, perhaps unparalleled in the evolution of MRI, resulting in dramatically improved imaging of MR angiography, cardiac function and cardiac perfusion. Our specific aims are to: 1) develop and evaluate improved methods to acquire, and reconstruct multislice non-Cartesian parallel imaging methods for 2D MRI applications 2) develop and evaluate robust combined non-Cartesian parallel imaging and compressed sensing methods 3) develop and evaluate improved computational methods based on graphical processing units (GPUs) for the calculation of non-Cartesian parallel imaging, CG-HYPR and combined methods to achieve clinically acceptable reconstruction times and 4) validate parallel CG-HYPR methods for the evaluation of cardiovascular disease as a means to shorten total exam time and increase image quality.