IMPROVED CINE MRI BY DIRECT-INVERSION RECONSTRUCTION

获取速度一直是电影磁共振成像（MRI）的关键问题。结构和

Acquisition speed remains a critical issue in cine magnetic resonance imaging (MRI). Structural and functional cardiac imaging, and increasingly also coronary artery and flow imaging, are becoming ever-more important tools in evaluation of a wide range of cardiovascular diseases. All of these scans often involve anatomy that is subject to both cardiac and respiratory motion. Multi-phase ECG-gated imaging is nowadays routinely performedin breath-hold mode, but both the duration and number of breath-hold acquisitions that can be tolerated by the patient are dominant limiting factorsfor image quality and resolution. These limitations are aggravated by the particular symptoms of many cardiac diseases. Faster MRI data acquisition is being pursued in multiple ways: improved gradient hardware allows faster spatial encoding; phased-array radiofrequency coils offer increased signal-to-noise ratio (SNR); new steady- state free precession acquisition techniques contribute improvements on both these fronts; other innovations, like partial-Fourier imaging, parallel imaging, or reduced field of view (rFOV) methods explore further improvements in imaging speed, often by trading some SNR, by exploiting various types of prior knowledge in the imaging model. Combined use of technologies that improve SNR or speed keeps pushing the limits of clinically feasible applications of cine imaging. However, not all techniques are always compatible, and the gains from all individual techniques are bounded, so the development of additional technologies in this realm is still desired and highly relevant. This proposal will investigate development and evaluation of a novel rFOV technique that promises to offer important advantages over comparable methods in terms of spatial and temporal resolution. The project builds on encouraging preliminary results from a retrospective implementation, using reduced subsets from conventionally acquired MRI data to obtain full-resolution cine reconstructions. Development is proposed of a full prospective implementation of the acquisition method on clinical equipment. Optimization will be investigated with respect to SNR, artifacts, and numerical efficiency. Qualitative and quantitative evaluations will be performed of image quality and flow calculations using the method. Finally, feasibility will be investigated of combining this method and parallel imaging methods.