基于中智相似积分算法的心肌超声造影分析系统自动评价心肌梗死大鼠左心室收缩功能  

Automatic evaluation of left ventricular systolic function in rats with myocardial infarction by myocardial contrast echocardiography based on neutrosophic similarity score algorithm

作　　者：郑雅朦 王飞[1] 张运帷 孙智超 王爽[1] 吴言[1] 郭延辉 田家玮[1] 杜国庆[1] Zheng Yameng;Wang Fei;Zhang Yunwei;Sun Zhichao;Wang Shuang;Wu Yan;Guo Yanhui. Tian Jiawei;Du Guoqing(Department of Ultrasound, the Second Affiliated Hospital of Harbin Medical University, the Key Laboratory of Myocardial Ischemia Ministry of Education （Harbin Medical University）, Harbin 150086, China)

机构地区：[1]哈尔滨医科大学附属第二医院超声科心肌缺血机理与诊疗技术省部共建教育部重点实验室(哈尔滨医科大学),150086 [2]Department of Computer Science, University of Illinois at Springfield, Springfield, IL USA

出　　处：《中华超声影像学杂志》2018年第5期434-440,共7页Chinese Journal of Ultrasonography

基　　金：国家自然科学基金（81671762,81371632）;黑龙江省博士后启动基金（LBH-Q16144）

摘　　要：目的探讨基于中智相似积分（NSS）算法的心肌超声造影（MCE）分析系统自动计算心肌梗死模型左室收缩功能的准确性及应用价值。方法SD大鼠制成心肌梗死模型，根据结扎位置高低分为较大梗死面积（MI-L）组和较小梗死面积（Mi-S）组，分别于术前、术后7d、术后28d行MCE检查。体外实验：基于NSS算法自动分割心内膜边界，并与人工描记计算机图像分割结果进行比较。体内实验：分别应用NSS的MCE分析系统、双平面Simpson法和心导管（PV—loop）计算左室射血分数（LVEF），获得NSS-LVEF、Simpson-LVEF和PV—LVEF；行Masson和HE染色计算心肌梗死面积。应用Bland—Altman评估三种LVEF计算法的一致性；计算组内相关系数（ICC）评价MCE分析系统的重复性。结果①MCE分析系统自动分割获得心内膜边界在三种算法中测值最小；②同组间三种方法测得LVEF差异无统计学意义（F=0．028，P=0．973），而MI-L与MI—S组间比较差异有统计学意义（F=78．61，P〈0．01）。NSS-LVEF与Simpson—LVEF和PV-LVEF均具有较好的一致性；③NSS算法观察者间及观察者内的相关系数（ICC）分别为0．96和0．98，说明NSS法的重复性良好；④与术前及术后7d相比，MI-L组和MI—S组术后28dLVEF均明显下降（P〈0．05），尤以MI-L组下降最为明显（P〈0．01）；且LVEF与心肌梗死面积呈明显负相关（r=-0．917，P〈0．01）。结论基于NSS算法的MCE分析系统可以实现对左室心内膜的自动准确识别，能较客观地评估实验大鼠左室收缩功能。Objective To evaluate the accuracy and value of automatically assessing left ventricular systolic function in rats with myocardial infarction （MI） by myocardial contrast echocardiography （MCE） based on neutrosophic similarity score（NSS） algorithm. Methods According to different infarction size （IS） ,SD rats were divided into large MI （MI-L, IS≥15%） and small MI （MI-S, IS〈15%） groups. MCE was performed before MI and at 7, 28 days after MI. In vitro study： the automatic segmentation of the endocardial contour based on neutrosophic similarity score algorithm was compared with the manual segmentation boundary. In vivo study： the left ventricular ejection fraction（LVEF） were calculated using the NSS system, biplane Simpson and PV-loop, respectively, and then NSS-LVEF, Simpson-LVEF and PV- LVEF were acquired, respectively. The IS was calculated by Masson and HE staining. The consistency between the two LVEFs was evaluated by Bland-Altman. The intraclass correlation coefficient （ICC） was calculated to evaluate the reproducibility of MCE analysis system. Results （1） The endocardial contour obtained by automatic segmentation had good agreement with the boundary of artificial tracing. （2）There was no significant difference among the three LVEFs in same group （ F = 0. 028, P = 0. 973）, but there was significant difference in different group （ F = 78.61, P 〈0.01）. NSS-LVEF was well consistent with both Simpson-LVEF and PV-LVEF. （3）The ICC of NSS algorithm for inter-observer and intra-observer were 0.96 and 0.98, respectively. （4）Compared with before MI and at 7 days after MI, the LVEF in MI-L and MI-S groups at 28 days after MI decreased significantly （ P 〈0.05）, especially in MI-L group （ P 〈0.01）. There was significantly negative correlation between LVEF and IS （r = -0.917, P 〈0.01）. Conclusions Left ventricular endocardium can be identificated automatically and LVEF can be calculated rapidly and accurately by MCE based on neutrosophic simi

关 键 词：超声心动描记术 心肌超声造影 心肌梗死 心室功能 左 图像处理 中智相似积分 计算机辅助 大鼠 

分 类 号：R445.1[医药卫生—影像医学与核医学] R542.22[医药卫生—诊断学]