基于直方图最小熵的MR平移运动伪影校正的逆向迭代新算法  

Inverse iterative correction for translational motion artifact of magnetic resonance imaging based on histogram entropy minimization

作　　者：江贵平[1] 陈武凡[1] 候振松[1] 

机构地区：[1]南方医科大学医学图像处理重点实验室,广东广州510515

出　　处：《第一军医大学学报》2005年第6期655-659,共5页Journal of First Military Medical University

基　　金：国家973项目(2003CB716100);国家自然科学基金重点项目(30130180);广东省自然科学基金项目(032888)~~

摘　　要：MR扫描过程中病人的运动使获得的重建图像含有运动伪影,严重影响临床诊断。本文针对MR图像刚性平移运动伪影提出了一个新的全自动的后处理方法:逆向迭代修正法(IIC)。利用本文提出的基于直方图的熵函数,IIC通过两个连续的步骤来消除模拟运动伪影:首先,在所有可能的运动方向上添加逆向相位差。然后,利用熵判断函数估计出实际的运动方向和位移,从而消除运动伪影。为了检验方法的可行性,IIC算法被用来消除由于刚性平移运动造成的MRI图像中的伪影。实验结果证明新方法与自动能量聚焦方法相比不仅在运行速度上有明显提高,而且能获得更好的校正图像。During the acquisition of a magnetic resonance images (MRI), blurring and ghosting artifacts caused by the patient's motion can seriously affect the result of diagnosis. A novel automatic post-processing strategy, inverse iterative correction (IIC), has been developed to suppress MRI artifacts due to the object's in-plane rigid-body motion. By means of the proposed histogram-based entropy function, IIC method uses two successive steps to reduce the simulated motion artifacts: first, the inverse phase errors are added to all possible simulated patient's motion directions, and in the second step, the actual directions and displacement from the patient's motion are estimated to properly correct the phase, hence remove the artifacts after searching all the trial directions. To verify its feasibility, the proposed method was used to reduce rigid-motion artifacts due to simulated motion in MRI images. The experimental results showed that the new algorithm significantly outperforms over the entropy auto-focus compensation algorithm on the quality of corrections for the motion artifacts and computational cost.

关 键 词：伪影校正 直方图最小熵 逆向迭代修正 刚性运动 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]