Propeller成像技术及图像重建算法的研究

发布时间：2018-04-27 17:39

本文选题：图像重建算法 + 运动伪影　；参考：《中南民族大学》2012年硕士论文

【摘要】：磁共振成像技术是生物学和医学领域研究的重要工具之一。目前磁共振成像已经广泛应用于临床，对人体疾病的诊断发挥着巨大作用。磁共振成像具有高分辨率、可任意方向断层成像以及无辐射等优点。随着成像技术的不断发展对成像质量和图像重建速度的要求越来越高。由于硬件水平、人体承受极限及梯度的切换速度使成像时间受到限制，致使病人常常会发生自主或非自主的运动导致成像中出现伪影，影响医生的诊断。如何消除运动伪影成为当今磁共振成像研究的技术热点。 PROPELLER成像技术可以减轻病人的运动伪影，缩短扫描时间，，该技术已经成功的应用在头颅磁共振成像中，对刚性运动伪影的消除效果显著。本文首先对磁共振成像技术的成像理论和图像重建方法进行了研究。深入分析了周期性旋转重叠平行线采集和增强后处理重建算法(Propeller)的原理，并对运动估计、非笛卡尔数据的重建算法等进行了深入的分析。其次对低场强下的Propeller图像重建算法进行实现，并对实现中的几个关键问题进行了深入的剖析。针对MRI采集的数据是非笛卡尔数据的特殊性，对算法中用到的几个关键运算的实现进行了研究，主要包括：原始数据文件的读取、显示K空间采样轨迹、运动估计、非笛卡尔数据的重建、密度补偿函数的选择等。图像重建采用的是Jackson网格化算法，随着采样数据的增多，对应的网格化的时间也就越长，密度补偿函数的选取也是难点。最后采用经典的Shepp-logan模型对propeller成像算法进行测试，模拟Propeller成像算法的K空间数据，运用propeller算法成功得到了重建图像。我们利用MATLAB来实现Propeller图像重建算法，该软件实现*.MRD文件的读取、数据处理、运动估计、密度补偿函数、网格化重建等进行研究分析，为深入研究消除运动伪影的方法及非笛卡尔数据的重建打下了基础。
[Abstract]:Magnetic resonance imaging (MRI) is one of the most important tools in biology and medicine. At present, magnetic resonance imaging has been widely used in clinical, and plays a great role in the diagnosis of human diseases. Magnetic resonance imaging has the advantages of high resolution, arbitrary direction tomography and no radiation. With the development of imaging technology, the quality of imaging and the speed of image reconstruction are becoming more and more important. Due to the level of hardware, the limit of human tolerance and the speed of gradient switching, the imaging time is limited, and the patient will often have spontaneous or involuntary motion, which will lead to artifacts in imaging, which will affect the diagnosis of doctors. How to eliminate motion artifacts has become the focus of magnetic resonance imaging. PROPELLER imaging technique can reduce the motion artifacts and shorten the scanning time. It has been successfully used in cranial magnetic resonance imaging and has a remarkable effect on the elimination of rigid motion artifacts. Firstly, the imaging theory and image reconstruction method of magnetic resonance imaging technology are studied in this paper. The principle of periodic rotation overlapping parallel line acquisition and enhancement post-processing reconstruction algorithm Propeller is deeply analyzed and the motion estimation and non-Cartesian data reconstruction algorithm are analyzed. Secondly, the Propeller image reconstruction algorithm under low field intensity is implemented, and several key problems in the implementation are deeply analyzed. In view of the particularity of non-Cartesian data collected by MRI, several key operations used in the algorithm are studied, including: reading of original data file, displaying K-space sampling trajectory, motion estimation, and so on. Reconstruction of non-Cartesian data, selection of density compensation function, etc. The Jackson mesh algorithm is used in image reconstruction. With the increase of sampling data, the corresponding time of gridding is longer, and the selection of density compensation function is also difficult. Finally, the classical Shepp-logan model is used to test the propeller imaging algorithm, and the K-space data of the Propeller imaging algorithm is simulated, and the reconstructed image is successfully obtained by using the propeller algorithm. We use MATLAB to realize the algorithm of Propeller image reconstruction. The software realizes the reading of Propeller files, data processing, motion estimation, density compensation function, mesh reconstruction and so on. It lays a foundation for further research on the method of eliminating motion artifacts and the reconstruction of non-Cartesian data.
【学位授予单位】：中南民族大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.0

【参考文献】