核磁共振成像(MRI)作为一种无创的检测手段在生物医学领域发挥着重要的作用。如何超快速获得高质量图像是MRI研究的重要课题。新近发展起来的时空编码成像技术对不均匀磁场和化学位移伪影有较好的抵抗性。本项目拟基于时空编码技术，研究能够同时抑制磁化率伪影、化学位移伪影、梯度涡流伪影、射频场不均匀伪影的超快速时空编码成像新方法。我们将研究时空编码成像中上述伪影的特点及其对自旋演化的共同作用，建立相应的物理模型，设计能够同时抵抗多种伪影的时空编码成像序列；开发无参考扫描畸变校正、水脂分离、基于深度卷积神经网络的伪影去除及超分辨率重建等算法，进一步改善成像图像质量；引入灵活视野、非线性采样、随机欠采样等高效采样技术提高三维成像效率；开展超快速定量磁化率成像、化学交换饱和转移成像等应用研究，从而建立一套能够在复杂非理想条件下超快速获得高保真图像的磁共振成像新技术，促进MRI在生物医学领域的应用。

Magnetic resonance imaging (MRI) is a non-invasive detection tool which plays an important role in biomedicine. How to ultrafast achieve high-quality MRI images is an important research topic. The newly-developed spatiotemporally encoded (SPEN) MRI technique has good immunity to inhomogeneous fields and chemical shift artifacts. In this project, based on the SPEN technique, we will explore novel ultrafast SPEN MRI methods capable of simultaneously suppressing susceptibility artifacts, chemical shift artifacts, eddy-current artifacts, and RF inhomogeneity artifacts. We will investigate the characteristics of abovementioned artifacts and their combined effects in SPEN MRI and establish related physical models. Based on the above work, we will design new SPEN MRI pulse sequences with exceptional resilience to these artifacts; we will develop referenceless image distortion correction algorithm, water-fat separation algorithm, deep convolutional network based artifact removal and super-resolved reconstruction algorithm, etc. to further improve the image quality; we will introduce efficient acquisition techniques like flexible field-of-view imaging, nonlinear acquisition and random undersampling to improve the efficiency of three-dimensional imaging; we will investigate the applications of new SPEN MRI methods on quantitative susceptibility mapping, chemical exchange saturation transfer imaging, etc. The new techniques we establish will be capable of ultrafast achieving high-fidelity images under complex nonideal conditions. The achievements would promote the application of MRI in biomedicine.