基于人工稀疏的动态磁共振成像算法研究

发布时间：2018-06-25 04:43

本文选题：动态磁共振 + 加速成像　；参考：《浙江大学》2017年硕士论文

【摘要】：时间分辨率和空间分辨率对于临床诊断动态磁共振成像来说具有非常重要的意义,因而促进了很多快速成像方法的产生。k-t principal component analysis(-t PCA)作为广泛使用的线性采集加速技术(κ-t broad-use linear acquisition speed-up technique,k-t BLAST)的延伸,是一种经典的高时间分辨率和高空间分辨率的重建算法,已经被广泛应用到自由呼吸的心脏灌注、脑灌注等具有高频带信号的动态快速成像中。它利用主成份分析技术提取时空相关关系作为时间基本函数,从根本上改变了信号重建逆问题,提高了重建图像的时间保真度。但是,当快速采样的加速因子不断提高时,k-tPCA的重建结果会随之出现不断增大的图像噪声和无法容忍的混叠伪影,极大地影响了 κ-tPCA算法在高帧频成像方面的应用。近年来,稀疏数据的重建已经应用在很多动态并行成像领域。通过将重建算法直接作用于稀疏性数据,可以得到比原始数据更好的重建结果。因此,针对k-t PCA的上述问题,本文进行了基于人工稀疏的动态磁共振成像算法研究。文中提出稀疏κ-tPCA重建算法,旨在结合人工稀疏的思想进一步提高k-t PCA的重建性能。通过消除由稀疏性重建得到的初始图像误差,提高重建的时间分辨率和信噪比,改善高倍加速重建中的图像质量。此外,基于现有的残差k空间稀疏性方法,本文又提出了残差k空间k-t PCA算法作为对比。文中通过仿真数据和实际人体成像数据分析了传统k-t PCA与这两种k-t PCA改进算法的重建结果,验证了理论方法的正确性,并利用均方误差、信噪比和信号强度时间曲线等指标分别描述重建算法的性能,在不同的欠采样数据中一一比较这几种算法的优劣,探索对于灌注成像等动态成像较为优秀的算法。
[Abstract]:Time resolution and spatial resolution are very important for clinical diagnosis of dynamic magnetic resonance imaging. As an extension of 魏 -t broad-use linear acquisition speed-up technique (魏 -t broad-use linear acquisition speed-up technique k-t blast), it is a classical reconstruction algorithm with high temporal resolution and high spatial resolution. It has been widely used in dynamic fast imaging of heart perfusion, cerebral perfusion and so on with high frequency band signal. It uses principal component analysis (PCA) technique to extract spatio-temporal correlation as a basic function of time, which fundamentally changes the inverse problem of signal reconstruction and improves the temporal fidelity of reconstructed images. However, when the acceleration factor of fast sampling is increasing, the image noise and the intolerable aliasing artifact will appear in the reconstruction result of k-tPCA, which greatly affects the application of 魏 -tPCA algorithm in high frame rate imaging. In recent years, sparse data reconstruction has been used in many dynamic parallel imaging fields. By applying the reconstruction algorithm directly to the sparse data, a better reconstruction result than the original data can be obtained. Therefore, a dynamic magnetic resonance imaging algorithm based on artificial sparsity is proposed to solve the above problems of k-t PCA. In this paper, a sparse 魏 -tPCA reconstruction algorithm is proposed to further improve the reconstruction performance of k-t PCA with the idea of artificial sparsity. By eliminating the initial image error obtained from sparse reconstruction, the time resolution and signal-to-noise ratio (SNR) of reconstruction are improved, and the image quality in high-power accelerated reconstruction is improved. In addition, based on the existing residual k-space sparsity method, the k-t PCA algorithm of residual k-space is proposed as a comparison. In this paper, the reconstruction results of the traditional k-t PCA and these two improved k-t PCA algorithms are analyzed by simulation data and actual human body imaging data. The correctness of the theoretical method is verified and the mean square error is used. The SNR and signal intensity time curves describe the performance of the reconstruction algorithm, compare the advantages and disadvantages of these algorithms one by one in different under-sampled data, and explore better algorithms for dynamic imaging such as perfusion imaging.
【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;R445.2

【相似文献】