基于K-TFOCUSS和自适应NLM3D的动态MRI重建  

Dynamic MRI Reconstruction Based on K-T FOCUSS and Adaptive NLM3D

作　　者：余永武[1] 郑迎凤[2] 

机构地区：[1]内江师范学院计算机科学学院,四川内江641000 [2]黄河科技学院现代教育技术中心,郑州450000

出　　处：《控制工程》2017年第4期768-774,共7页Control Engineering of China

基　　金：国家自然科学基金(71073033)

摘　　要：在动态磁共振成像中,采样时间过长既不利于对动态信息的获取也容易产生运动伪影。而传统的基于压缩感知(Compressive sensing,CS)的动态磁共振成像方法,具有重构算法慢,成像时间长等特点。基于此,将k-t FOCUSS(k-t Space Focal Underdetermined System Solver)和自适应非局部平均算法相结合应用于动态磁共振成像中,由k-t FOCUSS得到中等质量图像序列,再经过基于块的自适应非局部平均算法处理,最后得到的动态磁共振图像重建效果相较于单一的k-t FOCUSS重建效果更好。为了增加实验的对比性,还与k-t FOCUSS与ME/MC的结合算法做对比,经比较发现,提出的算法在保证重建图像质量的同时,计算时间更短。In dynamic magnetic resonance imaging, long sampling time is not conducive to acquire dynamic information and also prone to motion artifacts. However, the traditional method of dynamic magnetic resonance imaging based on compressive sensing is ineffective with slow reconstruction algorithm and long imaging time. Based on this, the combination of k-t FOCUSS and non-local adaptive averaging algorithm is applied to dynamic magnetic resonance imaging. After medium quality image sequence is obtained by k-t FOCUSS algorithm, the image sequence is processed through the non-local adaptive averaging algorithm based on block. Experiments demonstrate that the proposed method is better than the k-t FOCUSS reconstruction. To increase the contrast of the experiments, comparison with the algorithm of combining k-t FOCUSS and ME / MC is also done. The results show that the proposed algorithm ensures the reconstructed image quality and has a shorter computation time.

关 键 词：动态磁共振成像 压缩感知 k-t FOCUSS算法 基于块的自适应平均算法 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]