多相变分水平集的脑部磁共振图像分割与偏移场矫正的耦合模型

发布时间：2018-08-25 13:04

【摘要】：目的为分割有偏移场的脑部磁共振图像,建立了一种多相变分水平集的脑部磁共振图像分割与偏移场矫正的耦合模型。方法依据偏移场特性,定义了基于局部灰度信息的K均值聚类准则,并将聚类准则项整合为变分水平集函数的基本能量项,构造变分水平集的能量函数,求解能量函数的欧拉-拉格朗日方程。将提出的变分水平集算法扩展为多相,最终实现脑组织分割和偏移场的矫正。结果以医生手动分割的脑组织为标准,将本文模型与多相LBF模型的分割结果进行定量分析,结果表明,偏移场强度为20%时,本文模型的分割精度比多相LBF模型提高了10%,且随着偏移场强度进一步增强,本模型的优势更明显。结论本模型抗噪能力强且分割结果受参数影响小,可应用于脑组织的分割和偏移场的矫正。
[Abstract]:Aim in order to segment brain magnetic resonance images with offset field, a multi-phase change level set coupled model of brain magnetic resonance image segmentation and offset field correction is established. Methods based on the characteristics of migration field, the K-means clustering criterion based on local gray level information is defined, and the clustering criterion is integrated into the basic energy term of the variational level set function, and the energy function of the variational level set is constructed. The Euler-Lagrangian equation of energy function is solved. The proposed variational level set algorithm is extended to polyphase, and finally the segmentation of brain tissue and the correction of migration field are realized. Results the segmentation results of the present model and the multiphase LBF model were quantitatively analyzed according to the criteria of manual segmentation of brain tissue by doctors. The results showed that when the intensity of the migration field was 20, The segmentation accuracy of this model is 10% higher than that of multiphase LBF model, and the advantage of this model is more obvious with the further enhancement of migration field intensity. Conclusion the model has strong anti-noise ability and the segmentation results are less affected by the parameters, so it can be applied to the segmentation of brain tissue and the correction of deviation field.
【作者单位】：新乡医学院生物医学工程学院;
【基金】：国家自然科学基金(61305147) 新乡医学院科研培育基金(2014QN142)
【分类号】：TP391.41;R445.2

【参考文献】