|
|
自动化学报 2007
ECG信号自动诊断中回归建模法特征提取的研究DOI: 10.1360/aas-007-0462, PP. 462-466 Keywords: Autoregressivemodel,ECGfeatures,classification,automaticdiagnosis Abstract: ？Thisarticleexplorestheabilityofmultivariateautoregressivemodel(MAR)andscalarARmodeltoextractthefeaturesfromtwo-leadelectrocardiogramsignalsinordertoclassifycertaincardiacarrhythmias.TheclassificationperformanceoffourdifferentECGfeaturesetsbasedonthemodelcoefficientsareshown.Thedataintheanalysisincludingnormalsinusrhythm,atriaprematurecontraction,prematureventricularcontraction,ventriculartachycardia,ventricularfibrillationandsuperventriculartachycardiaisobtainedfromtheMIT-BIHdatabase.Theclassificationisperformedusingaquadraticdiscriminantfunction.TheresultsshowtheMARcoefficientsproducethebestresultsamongthefourECGrepresentationsandtheMARmodelingisausefulclassificationanddiagnosistool.
|
