基于广义级数的压缩感知磁共振并行成像方法研究

项目名称： 基于广义级数的压缩感知磁共振并行成像方法研究

项目编号： No.11301508

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 数理科学和化学

项目作者： 彭玺

作者单位： 中国科学院深圳先进技术研究院

项目金额： 22万元

中文摘要： 快速成像是磁共振用于病症临床诊断和治疗的关键，也是磁共振成像领域的技术难点。本项目针对目前快速成像方法的缺陷，提出基于广义级数、压缩感知和并行成像三者耦合的快速成像方法，克服广义级数模型所产生的系统误差，提高压缩感知技术所要求的信号稀疏性，改善并行成像系统的病态条件。本项目拟发展广义级数模型的自适应机制以提高模型对不同先验信息的鲁棒性，最大限度地改善压缩感知和并行成像技术；同时发展适于模型的稀疏变换设计方法，结合图像模型优化稀疏采样轨迹，以提高压缩感知技术在本方法中的性能。此外，基于快速自旋回波序列的原理，开发稀疏欠采样序列，直接获取真实欠采样数据并用于图像重建，探讨利用该方法加速磁共振成像的可行性与优缺点。本项目的实施，能够进一步丰富稀疏采样理论，推动快速磁共振成像技术取得新的突破，提高磁共振成像在重大疾病诊断中的临床应用能力和价值。

中文关键词： 磁共振成像；广义级数；压缩感知；T2弛豫成像；弥散张量成像

英文摘要： Fast imaging is the key of the utility of magnetic resonance imaging in desease diagnosis and clinic therapy, which is also one of the difficulties. To overcome the weakness of the state-of-the-art fast imaging methods, this project propose a generalizes series based method intergated with compressed sensing and parallel imaging, which can eliminate the system error caused by the generalized series model, enhance the signal sparsity requested by compressed sensing and alleviate the ill-conditioning of the parallel system. In this project, we will investigate the adaptability of the generalized series to various prior images in order to improve the behavior of compressed sensing and parallel imaging as much as possible; In the meanwhile, to further improve compressed sensing, a reference-driven sparsifying transform design scheme will be explored, sampling trajectory will be optimized according to the image model; Last but not the least, a sparse sampling sequence will be developed based on the standard turbo spin echo, such that direct under-sampled data can be acquired and be used in the proposed method for image reconstruction and method evaluation. The proposed work will further enrich the sparse sampling theory and may promote new breakthrough in realtime MRI and its clinical value in desease diagnosis.

英文关键词： magnetic resonance imaging；generalized series；compressed sensing；T2 relaxation mapping；diffusion tensor imaging