摘要
Prematureventricularcontraction(PVC)isthemostfrequentarrhythmiaencounteredinclinicalpractice.PVCmayoccurinhealthsubjects,whichisnotimminentlylife-threateningbutmayrequiretherapiestopreventfurtherproblems.So,thetimelyPVCrecognitionbecomesveryimportantfortheanalysisofelectrocardiogram(ECG),especiallyfortheremoteECGmonitoringusingmobilephones.Inthispaper,aconstructionmethodofpersonalizedECGtemplateandaPVCrecognitionmethodbasedontemplatematchingwerestudied.Firstly,weselected43ECGrecordingsfromtheMIT-BIHarrhythmiadatabase.Allrecordingsweredividedintotwodatasets(DS1fortrainingandDS2fortesting)andeachdatasetapproximatelycontainedthesameproportionofPVCbeats.Subsequently,foreachrecording(30min)inDS1,thefirst5minrecordingswereusedtoconstructthepersonalizedECGtemplateandthelast25minrecordingswereusedfortheR-wavepeaksdetectionandPVCrecognition,wherethetemplatematchingmethodwereused.ThevalidityoftheproposedmethodswastestedusingDS2.Theresultsshowedthat:1)highbeatdetectionaccuracywasachievedforbothPVCbeatsandnon-PVCbeats;2)thesensitivityandspecificityofPVCrecognitionwere99.11%and99.96%forthefirst5minrecordingsrespectively,99.17%and99.43%forthelast25minrecordingsrespectively.Alltheproposedmethodscanbereal-timeperformed,whichshowapromisingprospectfortheapplicationofECGmobilephones.
出版日期
2015年02月12日(中国期刊网平台首次上网日期,不代表论文的发表时间)