基于模拟退火优化的磁共振图像重建优化仿真  

Simulation on Optimization of Magnetic Resonance Image Reconstruction Based on Simulated Annealing Optimal Algorithm

作　　者：高志宇[1] 孙新娟[2] 

机构地区：[1]河南中医学院信息技术学院,河南郑州450046 [2]华北水利水电学院信息工程学院,河南郑州450011

出　　处：《计算机仿真》2013年第11期388-391,共4页Computer Simulation

摘　　要：研究医学磁共振成像的准确重建问题。不同器官的特征稀疏度差异较大,医学器官磁共振图像中包含了大量的曲线和边缘信息,造成重建图像不准确。传统的二维小波变换作为稀疏基来稀疏化医学图像,但是小波分解只限水平、垂直和对角三个方向,且变换滤波器是各向同性的,无法稀疏化地表示曲线状奇异特征,导致重建效果不好。提出了基于MRF参数约束的磁共振图像重建方法。建立了MRF模型与图像特征点参数的对应关系,采用模拟退火算法对模型求出最优解,得到重建参数和各特征点的准确坐标。仿真结果表明,改进算法能够较好的对磁共振图像的曲线和边缘特征进行合理的优化,使重建图像性能得到了改善和提高。The problem of precise reconstruction for the magnetic resonance image was researched in this paper. The sparsity of the different organs was great different, there were many curves and edge information in resonance im- ages, which resulte in the problem that the reconstruction of image was not precise. The 2-D wavelet transform was used as the sparse basis for processing the image traditionally, but the wavelet decomposition was limited to horizon- tal, vertical and diagonal direction. The transforming filter is of the same property that can not show the singular fea- ture of the curves sparsely. So the performance of the reconstruction is not satisfactory. An improved reconstruction method was proposed based on MRF restriction on the parameters. The relationship between the MRF model and fea- ture parameters was established, the optimal solution was reached based on the simulated annealing algorithm, and the accurate axis of the reconstructed parameters and feature points were obtained precisely. Simulation result shows that the curves and edge features are optimized reasonably, and the performance of the image reconstruction is im- proved greatly.

关 键 词：磁共振图像 图像重建 马尔科夫模型 

分 类 号：TP212.6[自动化与计算机技术—检测技术与自动化装置] TP391.9[自动化与计算机技术—控制科学与工程]