Objective To extract the contour of coronary formcoronary CT angiography image accurately and rapidly,we proposed a coronary artery segmentation model based on fuzzy clustering method and C-V model.Methods First,we made a coarse processing for the original image data.Then,the obtained membership matrix and the clustering information were coupled into the improved C-V model to complete the segmentation of the coronary artery image.Finally,qualitative and quantitative analysis of this model and the other two traditional models of coronary angiography image segmentation results were given.Results Qualitative analysis of the results:the improved model finished the coronary segmentation with fewer iterations.This model had a stronger ability to segment small and complex tissue,and the target edges were smooth.Quantitative analysis of the results:using the improved model iterated 200 times which took 11.722 s,overlapping rate was 83.42%;iterated 400 times which took 16.943 s,overlapping rate was 85.13%.Conclusions This model can finish the coronary segmentation with fewer iterations and have the characteristics of fast segmentation,strong anti-noise ability and smooth edge.It can be used to segment coronary,and provide a reference for the three-dimensional reconstruction of the image of the coronary.%目的 提出一种基于模糊聚类和改进C-V模型的新型图像分割方法,以精准和快速地提取冠状动脉CT血管造影图像中的冠脉轮廓.方法 首先对原始冠脉CT造影图像进行预处理;然后利用模糊C均值聚类算法进行预分割,将获得的隶属矩阵和聚类信息耦合进改进的C-V模型中,完成对冠脉图像的分割;最后定性和定量分析本文模型与其他两种传统模型对冠脉CT造影图像的分割结果.结果 定性分析结果显示,本文模型以较少的迭代次数完成了对冠脉轮廓的提取,对细小复杂的组织具有较强的分割能力,目标边缘光滑.定量分析结果显示,本文模型迭代200次耗时11.722 s、重叠率83.42%,迭代400次耗时16.493 s、重叠率85.13%.结论 结合模糊聚类的改进C-V模型能以较少迭代次数完成对冠脉轮廓的提取,具有分割速度快、抗噪能力强、目标边缘光滑等特点.该方法可以用于冠脉的分割,并为后续冠脉图像的三维重建研究提供参考.
展开▼
