磁共振线圈通道压缩算法在径向快速采集中的应用

发布时间：2019-05-28 14:59

【摘要】：磁共振并行成像技术利用多通道接收线圈空间灵敏度信息弥补空间编码不足,能够有效提升数据采集速度,改善图像信噪比。但是,随着接收线圈通道数量的不断增多,数据存储量以及计算复杂度也越来越大,尤其是在一些迭代计算中,因此数据处理和图像重建时间也随之增加。这对于磁共振三维采集、动态成像、实时成像以及快速成像而言,是一个亟需解决的问题。线圈通道压缩技术可以通过将多个接收线圈通道的原始数据通过线性拟合的方法得到较少的虚拟线圈通道,从而能够有效减少重建时间,同时没有丢失多通道接收线圈阵列的数据采集优势。目前,该技术在笛卡尔采样方式下已经得到了一定程度的应用,但是在非笛卡尔采样方式下的实现和应用却未见诸多报道。对此,本文首先研究和实现了磁共振线圈通道压缩技术在非笛卡尔采样方式下的应用,并以线性角径向采集和黄金角径向采集为例开展工作。本文主要内容包括:(1)研究和实现了基于线性角径向采集的GRAPPA(generalized auto-calibrating partially parallel acquisitions)和基于黄金角径向采集的 SPIRiT(iterative self-consistent parallel imaging reconstruction)的未采样数据恢复,并用网格化算法重建图像。(2)研究和实现了单一线圈通道压缩算法(single coil compression,SCC)和几何分解线圈通道压缩算法(geometric-decomposition coil compression,GCC)在径向快速采集中的应用,并在不同欠采样情况下对比两种算法重建所得图像与利用所有原始线圈通道数据重建所得图像的差异,计算重建时间分析算法效率,以及加入噪声后分析去噪能力。结果表明:SCC和GCC均能获得与所有原始线圈通道数据重建所得图像质量相当的结果,而在高倍欠采时更显优势,且在相同情况下GCC均优于SCC。SCC和GCC在去噪实验中也体现出更好的噪声容忍性,且GCC优于SCC。此外,SCC和GCC能有效减少重建时间,具有高效性。
[Abstract]:Magnetic resonance parallel imaging (MRI) can make up for the lack of spatial coding by using the spatial sensitivity information of multi-channel receiving coil, which can effectively improve the speed of data acquisition and the signal-to-noise ratio (SNR) of image. However, with the increasing number of receiving coil channels, the amount of data storage and computational complexity are increasing, especially in some iterative calculations, so the time of data processing and image reconstruction is also increasing. This is an urgent problem to be solved for three-dimensional magnetic resonance acquisition, dynamic imaging, real-time imaging and fast imaging. The coil channel compression technology can obtain less virtual coil channels by linear fitting of the original data of multiple receiving coil channels, thus effectively reducing the reconstruction time. At the same time, there is no data acquisition advantage of multi-channel receiving coil array. At present, this technique has been applied to a certain extent in Cartesian sampling mode, but the implementation and application in non-Cartesian sampling mode has not been reported. In this paper, the application of magnetic resonance coil channel compression technology in non-Cartesian sampling mode is studied and implemented, and the linear angular radial acquisition and gold angular radial acquisition are taken as examples to carry out the work. The main contents of this paper are as follows: (1) the unsampled data recovery based on linear angle radial acquisition GRAPPA (generalized auto-calibrating partially parallel acquisitions) and gold angle radial acquisition SPIRiT (iterative self-consistent parallel imaging reconstruction) is studied and implemented. The image is reconstructed by gridding algorithm. (2) the application of single coil channel compression algorithm (single coil compression,SCC) and geometric decomposition coil channel compression algorithm (geometric-decomposition coil compression,GCC) in radial fast acquisition is studied and implemented. The differences between the images reconstructed by the two algorithms and those obtained by using all the original coil channel data are compared under different undersampling conditions, and the efficiency of the reconstruction time analysis algorithm and the denoising ability after adding noise are calculated. The results show that both SCC and GCC can obtain the same image quality as all the original coil channel data, and have more advantages in the case of high power undermining. In the same case, GCC is superior to SCC.SCC and GCC in denoising experiment, and GCC is superior to SCC. in noise tolerance. In addition, SCC and GCC can effectively reduce reconstruction time and are efficient.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;R318

【相似文献】