摘要
目的为便捷有效地识别肥心病心音与正常心音,针对心音信号研究基于WER-PCA的肥心病心杂音特征提取算法。方法采集168例正常心音信号和194例肥心病心音信号,应用小波变换获得10维重构信号,应用主成分分析筛选出肥心病心杂音频段对应的小波分解特定层,提取心杂音频段的能量百分比,将其与正常心音的差异度相组合作为肥心病心音特征。结果对比正常心音与肥心病心音主成分构成元素,筛选出肥心病心杂音频带范围为86.15~689.05Hz;对正常心音和肥心病心音进行识别,差异度结合能量的识别特征,其最优识别正确率为95.3%。结论该算法能有效提取肥心病心杂音特征以识别肥心病。
Objective To identify the hypertrophic cardiomyopathy(HCM)heart sound and normal heart sound conveniently and effectively,the WER-PCA-based method for extracting heart murmur features of HCM heart sounds was studied.Methods One hundred and sixty eight normal heart sounds and 194 HCM heart sounds were collected.Wavelet transform was used to obtain the 10-dimensional reconstructed signal and principal component analysis(PCA)was used to select the murmur segment of the HCM corresponding to the specific layer of wavelet decomposition.In addition,the wavelet energy percentage of the murmur segment was extracted.The combination of the wavelet energy percentage of the murmur segment in HCM with the difference between the HCM heart sound and the normal heart sound was used as the identification feature of HCM.Results Comparing the main components of normal heart sounds and HCM heart sound,the frequency range of HCM heart sounds was from 86.15 Hz to 689.05 Hz.The comprehensive correct rate of heart sound recognition in normal heart sounds and HCM heart sounds was 95.3%.Conclusion The algorithm can effectively extract heart murmur features of HCM for identification.
作者
张小兰
房玉
刘栋博
王维博
王海滨
Zhang Xiaolan;Fang Yu;Liu Dongbo;Wang Weibo;Wang Haibin(School of Electrical and Electronic Information,Xihua University,Chengdu Sichuan 610039,China)
出处
《航天医学与医学工程》
CAS
CSCD
北大核心
2020年第1期59-65,共7页
Space Medicine & Medical Engineering
基金
国家自然科学基金(61571371)
西华大学研究生创新基金(ycjj2019052)
西华大学大健康管理促进中心2019年开放课题(DJKG2019-008
DJKG2019-005).
关键词
肥厚型心肌病
心音
小波变换
主成分分析
能量百分比
hypertrophic cardiomyopathy(HCM)
heart sound
wavelet transform
principal component analysis
energy percentage
