MR图像运动伪影的遗传算法校正  

Motion Artifact Correction in MRI Using Genetic Algorithms

作　　者：陈春晓[1] 陶华[1] 王世杰[1] 罗立民[1] 

机构地区：[1]东南大学生物医学工程系

出　　处：《中国图象图形学报》2005年第9期1129-1133,共5页Journal of Image and Graphics

基　　金：国家重点基础研究发展规划"973"项目(2003CB716102)

摘　　要：磁共振成像过程中,由于患者的生理运动或自主性运动会使扫描数据发生相位偏移,致使重建图像含有运动伪影,从而降低成像的质量,而且严重的伪影则会影响医生对病灶的精确定位。为了能有效地抑制运动伪影,利用遗传算法高度并行、随机和自适应全局寻优的特点,提出了基于遗传算法的运动伪影修正方法,即采用逐次修正扫描信号在K空间中已偏移的相位的方法来抑制伪影。基于遗传算法的修正方法,由于能有效校正扫描数据的相位偏移,因而可达到抑制运动伪影的目的。实验表明,该方法对由较小运动引起的伪影有很好的抑制作用,对含噪声及较大运动引起的伪影的抑制作用,与经典的迭代算法相比,也有很大地改善。Motion artifact suppression is a very difficult problem in magnetic resonance imaging. Patient motion including physiological motion and physical motion causes the phase distortion in the collected signals and induces motion artifacts in the reconstructed image for the 2-dimension Fourier Transform imaging method. As the result of these motion artifacts, the quality of image is degraded and the precise orientation to the focus in clinic is affected. A motion artifact reduction method based on genetic algorithm is presented in this paper. Genetic algorithm has the characteristics of parallel, randomicity and adaptive matching pursuit. In the image reconstruction, before taking the inverse Fourier Transform, the phase distortion of K-space signals is compensated step by step through searching for the optimizing phase values. The experiments show that the phase distortion can be estimated using the information implied by the motion artifacts and a significant amount of motion artifact suppression is achieved. Using this algorithm, the corrected image is satisfied when motion is slight, and the quality of the image is still improved greatly in the conditions of noise and significant motion comparing with using classical iterative algorithms.

关 键 词：磁共振成像 运动伪影 遗传算法 

分 类 号：R445.39[医药卫生—影像医学与核医学] TP391.4[医药卫生—诊断学]