超声背散射统计参数成像评价儿童脂肪肝的新方法  

New Method for Evaluating Fatty Liver in Children by Ultrasound Backscattering Statistical Parameter Imaging

作　　者：党英楠 高瑞阳 宾光宇[1] 吴水才[1] 崔博翔 周著黄[1] DANG Yingnan;GAO Ruiyang;BIN Guangyu;WU Shuicai;CUI Boxiang;ZHOU Zhuhuang(Beijing International Science and Technology Cooperation Base for Intelligent Physiological Measurement and Clinical Transformation,Faculty of Environment and Life,Beijing University of Technology,Beijing 100124,China;College of Medicine,Chang Gung University,Taoyuan Taiwan 333323,China)

机构地区：[1]北京工业大学环境与生命学部,智能化生理测量与临床转化北京市国际科研合作基地,北京100124 [2]台湾长庚大学医学院,中国台湾桃园333323

出　　处：《中国医疗设备》2023年第9期17-24,共8页China Medical Devices

基　　金：北京市自然科学基金(4222001);国家自然科学基金(11804013;61871005)。

摘　　要：目的超声零差K分布是最具物理意义的背散射包络统计模型,探讨超声零差K成像评价儿童脂肪肝的可行性。方法改进基于神经网络的零差K分布参数估算方法,采用特征值的理论值训练估算模型,将并行计算引入背散射统计参数成像。利用中心频率3.5 MHz的探头采集儿童的背散射信号,将改进后的背散射统计参数成像应用于儿童脂肪肝的评价,分析Nakagami成像m参数、零差K成像k参数和log10(α)参数诊断儿童脂肪肝的性能,包括3种二分类(G0 vs.G1~G3、G0~G1vs.G2~G3、G0~G2vs.G3)。结果共得到393例儿童脂肪肝超声背散射信号,其中正常G0共165例,轻度G1共33例,中度G2共78例,重度G3共117例。并行计算法的超声Nakagami成像和超声零差K成像的平均运行时间均比滑动窗口法明显缩短,并行计算法的运算速度约是滑动窗口法的3倍。log10(α)参数在3种二分类中,均获得了最高的曲线下面积、特异性和准确度。结论本研究设计基于并行计算的超声背散射统计参数成像方法,可提高成像速度,超声背散射统计参数成像能用于儿童脂肪肝的定量评价,特别是早期检测,研究结果可为儿童脂肪肝的定量超声评价提供新思路和新方法。Objective Ultrasonic homodyne K distribution is a statistical model of backscattering envelope with the most physical parameters.To explore the feasibility of ultrasonic homodyne K imaging in evaluating children’s fatty liver.Methods An improved method of homodyne K distribution parameter estimation based on neural network was proposed to train the estimation model with eigenvalue,and parallel computation was introduced into backscattering statistical parameter imaging.The backscattering signals of children were collected with a 3.5 MHz probe.The improved backscattering statistical parameter imaging was applied to the evaluation of fatty liver in children,and performance of the Nakagami imaging m parameter and the homodyne K imaging k parameter and log10(α)parameter in diagnosing children’s fatty liver was analyzed.Three classifications were performed:G0 vs.G1-G3,G0-G1 vs.G2-G3,G0–G2 vs.G3.Results A total of 393 cases of fatty liver ultrasound backscattering signals were obtained,of which 165 cases were normal G0,33 cases were mild G1,78 cases were moderate G2 and 117 cases were severe G3.The average running time of the parallel computing method for ultrasonic Nakagami imaging and ultrasonic homodyne K imaging was significantly shorter than that of the sliding window method,and the computing speed of the parallel computing method was about 3 times that of the sliding window method.The parameters of log10(α)obtained the highest area under the curve,specificity,and accuracy in all three binary classifications.Conclusion In this study,an ultrasound backscattering statistical parameter imaging method based on parallel computing is designed to improve the imaging speed.Ultrasound backscattering statistical parameter imaging can be used for quantitative evaluation of children’s fatty liver,especially for early detection,and the research results can provide a new idea and a new method for quantitative ultrasound evaluation of children’s fatty liver.

关 键 词：超声背散射 超声零差K分布 儿童脂肪肝 包络统计模型 并行计算法 

分 类 号：R197.39[医药卫生—卫生事业管理] TP391[医药卫生—公共卫生与预防医学]