基于高阶和非局部全变分的核磁共振图像重构算法  

Higher degree-and non-local total variation based MRI reconstruction algorithm

作　　者：陈超楠[1] 杨晓梅[1] 

机构地区：[1]四川大学电气信息学院,成都610065

出　　处：《计算机应用研究》2016年第6期1903-1906,1912,共5页Application Research of Computers

摘　　要：为了更好地提高核磁共振图像(magnetic resonance image,MRI)重构质量,提出了一种基于高阶和非局部全变分的混合重构算法。首先,该算法采用全局梯度提取法分离待重构MRI图像的平滑区和边缘纹理区;在算法的每次迭代过程中,先对平滑区采用高阶全变分(higher degree total variation,HDTV)算法,然后对边缘纹理区采用非局部全变分(non-local total variation,NLTV)算法;最后,将算法迭代结束后的平滑区和边缘纹理区图像合并就是重构图像。实验发现,该算法的重构效果大大优于TV(total variation,全变分)、HDTV、NLTV算法,其重构图像既能有效地滤除噪声和保留纹理细节信息,又大大抑制了全变分的阶梯效应。In order to improve magnetic resonance image（ MRI） reconstruction quality,this paper proposed a hybrid reconstruction algorithm,which combined higher degree total variation（ HDTV） and non-local total variation（ NLTV）. First,it decomposed the reconstructed MRI into constant area and detail area by a global gradient extraction scheme. Second,in each iteration,the hybrid reconstruction algorithm applied HDTV on the constant area firstly,then it applied the NLTV on the detail area. At last,the MRI reconstruction image was combining the constant area with detail area at the end of iteration. The experiments demonstrate that the proposed algorithm is superior to total variation algorithm,HDTV and NLTV. It not only removes noise effectively and preserves the texture and detail information better,but also greatly suppresses the staircase of the total variation.

关 键 词：核磁共振图像 高阶全变分 非局部全变分 平滑区 边缘纹理区 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]