基于增强补种的混合水平集 CT 图像分割

摘要

针对医学图像 CT 图像像素不均匀对图像局部分割算法影响较大的问题，提出一种基于 La-grangian 粒子增强补种算法的混合水平集医学 CT 图像分割算法。首先，针对局部图像的非均匀性，通过在计算水平集公式前先计算 Lagrangian 标记粒子来重建内嵌交接界面，从而提高水平集算法的质量守恒特性；其次，针对传统粒子方法在处理界面奇异性和复杂几何相关问题上的不确定性，通过增加速度矢量和单位法向量来促进奇异点和拓扑变化点速度场的收敛；最后，通过在合成数据测试集和真实 CT 图像上的仿真测试表明，所提算法在边缘分割收敛精度及运算速度上均要优于对比算法。%According to the problem of pixel nonuniform on local medical CT image that affecting the segmentation algo -rithm ,lagrangian particle reseeding based mixed level set algorithm for medicine CT image segmentation was proposed . Firstly ,according to the non uniformity for local image ,lagrangian marker particles were calculated before executing the lev-el set algorithm to reconstruct the embedded interface ,whichimprove the mass conservation properties of the level set algo-rithm ;Secondly ,in order to address the uncertainty of interface singular and complex geometry relevance ,the velocity vec-tor and the unit normal vector were used to promote the convergence of velocity field for singular point and topology change point ;Finally ,through the simulation test in the synthetic data test set and real CT images show that the proposed algo-rithm in edge segmentation accuracy and speed of operation is better than comparison algorithm .