用纹理分析的方法实时量化超声粒子图像测速技术中微泡的浓.pdfVIP

用纹理分析的方法实时量化超声粒子图像测速技术中微泡的浓度 用纹理分析的方法实时量化超声粒子图像 测速技术中微泡的浓度 Real-Time Quantification of Microbubbles Concentration for Ultrasonic Particle Image Velocimetry Technique Using Texture Analysis Method 牛丽丽 钱明 靳巧峰 姜波 郑海荣 摘 要 之前的研究表明在成像过程中超声造影剂的浓度影响超声粒子图像测速技术的计算精度。这篇文章陈述了一种基于 纹理的方法来获得超声粒子图像测速过程中最优的造影剂浓度。分别用10MHz和30MHz的超声系统对注射不同浓度造影剂 的体外旋转流场和体内大鼠颈总动脉流成像(B超图像)，然后一阶纹理特征和二阶纹理特征(标准偏差、对比度、能量、一致 3 5 性) 被获得从这些图像。结果表明最优的造影剂浓度在体外是0.8~2×10 bubbles/ml ，在体内是1~5×10 bubbles/ml； 而且在这个浓度范围内，应用超声粒子图像测速技术可获得较高精度的速度矢量图。在这个浓度范围外，获得的速度矢量图 产生很多错误。体外实验应用最近邻居分类法分类精度可达到86.4%。因此，纹理分析方法可用来实时的获得最优的造影剂 浓度在超声粒子图像测速过程中。 关键词 超声造影剂；超声粒子图像测速；纹理分析 ABSTRACT　Previous studies showed that the concentration of ultrasound contrast agents (UCAs) during imaging would impact greatly on the accuracy of flow velocity measurements using ultrasonic particle image velocimetry (Echo PIV) technique. This paper presents a texture-based method to identify the optimum concentration of UCAs for Echo PIV. The first-order and seconded-order texture features, standard deviation, contrast, energy and homogeneity, are extracted from ultrasound images of in vitro rotational flow (10MHz) and in vivo rat (30MHz) common carotid arterial flow with UCAs at various concentrations. The results show that the optimum UCAs concentration is 0.8~2×103 bubbles/ml in vitro and 1~5×105 bubbles/ml in vivo, therein high accurate velocity vectors are obtained by Echo PIV. Otherwise, poor quality velocity vectors are yielded. The accuracy of K-nearest neighbor classifier reaches 86.4% in vitro. Thus, texture features could be used to identify the optimum UCAs concentration in real time for Echo PIV. KEYWORDS UCAs; Echo PIV; Texture analysis 1 Introduction