IDENTIFICATION OF PVC BEATS BY NEURAL NETS

Chikh, M.a; Belgacem, N; Meghnefi, F; Bereksi-reguig, F

Sciences & technologie. B, Sciences de l’ingénieur
Volume 0, Numéro 20, Pages 28-32
2003-12-31

Identification Of Pvc Beats By Neural Nets

Authors : Chikh M.a . Belgacem N . Meghnefi F . Bereksi-reguig F .

Abstract

This paper describes the design, training and testing of an artificial neural network for classification of normal and abnormal premature ventricular contraction (PVC) beats in ECG signal. To carry out the classification task, we use the back-propagation (BP) learning algorithm. Two feature selections types were investigated with aim of generating the most appropriate input vector for the artificial neural network classifier (ANNC). The first selected information of each ECG beat is stored as 33-element vector; the second one is then reduced to a 10 dimensional vector using principal component analysis (P.C.A). The performance measures of the classifier will also be presented using as training and testing data sets from the MIT-BIH database.