一种基于淘汰粒子群优化聚类算法的心肌超声造影分析系统

A quantitative analysis system for myocardial contrast echocardiography based on eliminating particle swarm optimization clustering algorithm

背景:心肌超声造影已成为评价心肌微灌注有效的影像学方法之一,但目前心肌超声造影图像的分析判断多采用目测法或手动描记感兴趣区域,很大程度上依赖观察者的经验,缺乏客观定量分析方法。目的:探讨新型心肌超声造影图像定量分析系统评价心肌灌注的可行性。方法:采集家兔心肌超声造影图像,应用基于淘汰粒子群优化聚类(EPSO)算法的计算机图像定量分析系统对造影图像进行自动处理,获得心肌灌注定量参数。结果与结论:①图像识别软件自动识别心内膜与心外膜,确定心肌区域。②自动将心肌区域分割为相等的6个区。③自动标记各节段心肌的标化对比剂密度值。④以标化对比剂密度值为依据行彩色编码,获得彩色编码面积百分比。⑤自动生成心肌节段灌注分布直方图。基于淘汰粒子群优化聚类算法的心肌超声图像分析软件,在定量评价心肌微灌注和识别灌注异常方面具有较好的可行性和较高的应用价值。

BACKGROUND：Myocardial contrast echocardiography（MCE） has been used to assess the myocardial capillary perfusion after thrombolytic therapy or percutaneous coronary intervention.The current interpretation of perfusion images by MCE is performed by visual qualitative analysis and drawn manually the region of interests.The accuracy of such analysis depends on the investigators＇ experience,and lacks quantitative analysis.OBJECTIVE：To evaluate the feasibility of a quantitative analysis system for MCE in assessing myocardial perfusion.METHODS：Rabbit MCE images were collected,and processed automatically by quantitative analysis software based on eliminating particle swarm optimization（EPSO） clustering algorithm to obtain myocardial perfusion parameters.RESULTS AND CONCLUSION：①Quantitative analysis software was used to identify automatically endocardium and epicardium and determine myocardial region;②Six myocardial regions were dissected automatically;③Calibrated myocardial contrast intensity（CI） of all segments was calculated automatically;④Myocardial perfusion of all segments was color-coded according to calibrated CI,and the percentage of color-coded area was obtained;⑤Myocardial perfusion distribution histogram was generated automatically.MCE image analysis system based on EPSO clustering algorithm in the quantitative assessment of myocardial microperfusion and identification of myocardial perfusion abnormalities has good feasibility and high value.