论文部分内容阅读
为了更好地利用相位信息补偿血流造成的影响,本文对磁敏感加权成像(SWI)中存在的相位缠绕问题展开了研究。为提高解缠绕的准确性,本文提出了幅度图像引导的磁敏感加权图像相位解缠算法。基本思路如下:(1)通过改进旋转不变非局部主成分分析滤波(PRI-NL-PCA)降低噪声影响;(2)结合C-V模型水平集提取相位图像中对应的实性组织区域,从而规避背景噪声对相位解缠方法的影响;(3)采用相位补偿的方法约束K空间重建出的相位图像。最后,利用四种统计量作为量化指标,评价解缠绕方法的可靠性:相位误差的突变点个数、均值(M)、方差(Var),以及阳性百分比(Pos)和阴性百分比(Neg)。通过对比仿真数据和226组真实头部磁敏感数据,结果表明,本文算法相对于经典的枝切法和最小二乘法,解缠绕结果具有较高的准确性。
In order to make better use of the phase information to compensate for the influence of blood flow, the paper studies the problem of phase winding existing in magnetically susceptible weighted imaging (SWI). In order to improve the accuracy of unwrapping, this paper presents an amplitude-image-guided phase unwrapping algorithm for magnetically susceptible weighted images. The basic idea is as follows: (1) The PRI-NL-PCA is improved to reduce the influence of noise; (2) the real solid tissue area in the phase image is extracted by the CV model level set to avoid The effect of background noise on the phase unwrapping method; (3) phase compensation method is used to restrain the phase image reconstructed from K space. Finally, the reliability of the unwrapping method is evaluated using four statistics as a quantitative indicator: the number of mutation points, the mean (M), the variance (Var), and the Pos and Neg values of the phase error. By comparing the simulated data with 226 sets of real head magnetosensitive data, the results show that the proposed algorithm has higher accuracy than the traditional branch and least square method.