中文摘要：

定量磁化率成像（quantitative susceptibility mapping，QSM）技术大多采用多回波梯度回波序列采集相位数据，经加权最小二乘法（weighted linear least-square，WLS）拟合得到局部磁场分布.对于组织磁化率分布不均匀的区域，尤其是颅底部位，常规WLS算法拟合得到的局部磁场误差较大，导致相应部位磁化率分布图信噪比较低.针对常规WLS算法的这一不足，该文提出了一种截断WLS算法.对两种算法拟合得到的磁化率分布图对比研究发现，截断WLS算法可有效提高颅底部位定量磁化率分布图的图像质量，使其噪声明显下降.

英文摘要：

A weighted linear least-square （WLS） algorithm is generally applied for multi-echo phase data to estimate the voxel-by-voxel field shift, which is further used for inversion of susceptibility map. The fitting error on the estimated field map, induced from conventional WLS method, may lead to artifacts and low signal-to-noise ratio on the susceptibility map, especially in the regions with inhomogeneous distribution of magnetic susceptibility. To improve the accuracy of the estimated field map, a truncated WLS approach was used to truncate the signal of low signal-to-noise ratio and capture the field information before the signal in a voxel decays to the noise level, which can enhance the contrast of structures in the bottom of the brain on the susceptibility map. Experimental studies demonstrated that susceptibility noise was dramatically reduced by method of truncated WLS.