实时三维时空编码磁共振成像的序列设计及重建算法研究

Real-time three-dimensional (3D) magnetic resonance imaging (MRI) can continuously monitor various physiological activities and has many important clinical applications. However, technical challenges remain in many aspects, such as low spatio-temporal resolution and poor robustness. Standing as one of MRI’s cornerstones is echo-planar-imaging (EPI), Sir Peter Mansfield’s Nobel-award winning proposition for the collection of two-dimensional (2D) MRI scans in a single shot. Despite its central role as the basis of fast imagingtechnologies, EPI faces limitations in real-time 3D MRI. The foremost challenge is the difficulty in dealing with distortions associated to magnetic field heterogeneities in inhomogeneous tissues, as well as with artifacts due to fat signal. Over the last few years an alternative “ultrafast” 2D scanning technique, based on so-called SPatio-temporal ENconding (SPEN) principles were introduced and advanced. Unlike EPI, SPEN MRI is not based on Fourier transformation and thereby is not constrained by Nyquist criteria. It can provide images with unprecedented robustness and higher internal spatial-resolution. However, in its present implementation, SPEN is still not fast enough for real-time 3D imaging. The higher specific absorption rate (SAR) of 3D SPEN is another obstacle for its clinical applications. The objective of this proposal is to develop new, more efficient methods for real-time 3D MRI. We will improve SPEN scheme for better spatio-temporal resolution, signal-to-noise ratio and robustness based on super-resolution reconstruction, compressed sensing, field correction algorithms, spiral sampling and segmented spatiotemporal encoding, etc. which we have been developed over the last years. The new SPEN methods would enable real-time 3D MRI studies on humans accessing a variety of physiological processes such as heartbeat, motion of articulations and digestive/respiratory processes. It would also be applied in diffusion and functional MRI, perfusion MRI and chemical exchange saturation transfer studies at high fields.

实时三维磁共振成像(MRI)可以实时监控各种生理活动，具有重要的临床应用价值。然而该技术在时-空分辨率及可靠性等方面仍存在许多挑战。曾获得过诺贝尔奖的回波平面成像(EPI)方法作为快速MRI的一个基石，因易受不均匀场或化学位移伪影影响而不适用于多数实时三维MRI领域。近年来出现的单扫描二维时空编码(SPEN)相比EPI具有更高可靠性及内在分辨率。然而，二维SPEN技术要推广到实时三维MRI仍然存在一系列技术障碍，如时-空分辨率仍然无法满足许多实时三维MRI的要求，较高的SAR值无法满足人体成像的安全要求等。本项目拟利用我们近几年开发的基于超分辨率重建、压缩感知、畸变校正、螺旋采样、分段编码等技术提升SPEN的时-空分辨率、信噪比和可靠性，建立一套可应用于心脏、关节和消化/呼吸系统等各种生理活动的三维实时人体成像的新方法，并应用到高场下的扩散、脑功能、灌注、化学交换等成像领域。

实时三维磁共振成像可以实时监控各种生理活动，快速获取多种对比度成像，具有重要的临床应用价值。然而该技术在时-空分辨率及可靠性等方面仍存在许多挑战。曾获得过诺贝尔奖的回波平面成像(EPI)方法作为快速MRI的一个基石，因易受不均匀场或化学位移伪影影响而不适用于多数实时三维MRI领域。近年来出现的单扫描二维时空编码(SPEN)相比EPI具有更高可靠性及内在分辨率。然而，二维SPEN技术要推广到实时三维磁共振成像仍然存在一系列技术障碍，如时-空分辨率仍然无法满足许多实时三维MRI的要求，多种伪影的干扰仍然限制成像的质量，较高的SAR值无法满足人体成像的安全要求等。本项目利用我们近几年开发的基于超分辨率重建、压缩感知、畸变校正、螺旋采样、分段编码等技术提升SPEN的时-空分辨率、信噪比和可靠性，建立一套可应用于多个成像区域的三维实时成像的新方法，并应用到高场下的扩散、脑功能、灌注、化学交换等成像领域。在本项目的四年研究期内，我们取得如下几个方面的重要成果：（1）我们开发设计了新型的时空编码超快速成像序列，并引入分段编码技术以解决三维超快速时空编码成像序列的高SAR值的问题，我们还引入空间域直接螺旋采样、多重聚回波及多重叠回波技术进一步提升时空编码成像序列的采集速度，实现了三维实时成像的目标；（2）我们率先在时空编码成像领域引入了深度学习超分辨重建技术，进一步提升时空编码成像的重建质量，同时获得优秀的无参考扫描扭曲校正效果；（3）在项目资助下，我们进一步将所发明的一系列超快速成像技术应用到多个不同的成像领域，包括扩散成像、化学交换成像及水脂分离成像等，并取得许多重要的成果。本项目资助下的相关研究成果已经在国际主流SCI刊物发表论文16篇，并获得中国发明专利4项授权，更多的相关成果仍然在整理中，并将于近期投稿国际权威刊物。