应用小波多尺度特征检测机械通气人机不同步  

Application of Wavelet Multi-scale Characteristics to Detect Patient-ventilator Asynchrony in Mechanical Ventilation

作　　者：陆云飞 陆飞[1] 方路平[1] 葛慧青[2] 潘清[1] LU Yun-fei;LU Fei;FANG Lu-ping;GE Hui-qing;PAN Qing(College of Information Engineering,Zhejiang University of Technology,Hangzhou 310023,China;Department of Respiratory Care,Sir Run Run Shaw Hospital,School of Medicine,Zhejiang University,Hangzhou 310016,China)

机构地区：[1]浙江工业大学信息工程学院,杭州310023 [2]浙江大学医学院附属邵逸夫医院呼吸治疗科,杭州310016

出　　处：《小型微型计算机系统》2020年第12期2677-2682,共6页Journal of Chinese Computer Systems

基　　金：浙江省自然科学基金项目(LY19H010005)资助;浙江省教育厅一般项目(Y201636066)资助。

摘　　要：机械通气人机不同步检测对于改善重症病人的呼吸状态具有重要意义.本文提出了一种应用小波多尺度特征的算法对其进行自动检测.首先对原始呼吸波形进行预处理;然后在对呼吸波形进行离散小波变换的基础上,根据不同步波形序列波动的不规律性和不可预测性的特点,提出了一种新的特征提取方法,即对不同层次的信号提取多种熵值特征以及非线性特征;利用序列前项选择算法选择最佳特征;最后利用SVM得到人机不同步分类结果.将算法应用于临床采集的数据集,结果显示,利用所选择特征分类效果理想,灵敏度和特异性达到了93.41%和96.68%,优于传统方法,有望辅助重症机械通气患者的治疗.Patient-ventilator asynchrony detection of mechanical ventilation has an important meaning to improve the respiratory status of critically ill patients,This paper proposed an algorithm based on multi-scale features of wavelet to detect it automatically.First,the original breathing wave was preprocessed.Then,based on the discrete wavelet transform of the breathing waveform,a new feature extraction method is proposed according to the regularity and unpredictability of the wave sequence of the asynchrony waveform,that is,multiple entropy and non-linear features are extracted from signals at different levels.The method used the sequential forward selection algorithm to select the optimal features.Finally,the method used SVM to obtain the results of patient-ventilator asynchrony classification.The results showed that the selected feature classification was effective,the sensitivity and specificity of the proposed algorithm reached 93.41%and 96.68%,w hich were better than traditional methods and it is expected to assist the treatment of patients with severe mechanical ventilation.

关 键 词：离散小波变换 机械通气 人机不同步 特征选择 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]