NOVEL ELECTROCARDIOGRAPHIC MARKERS BASED ON WAVELET TRANSFORM FOR PREDICTION OF VENTRICULAR ARRHYTHMIAS
-
Published:2010-12
Issue:06
Volume:22
Page:481-488
-
ISSN:1016-2372
-
Container-title:Biomedical Engineering: Applications, Basis and Communications
-
language:en
-
Short-container-title:Biomed. Eng. Appl. Basis Commun.
Author:
Uchaipichat Nopadol1,
Sinsomboonchai Krittaya1,
Jongworakun Kuncharee1
Affiliation:
1. Department of Electrical and Computer Engineering, Faculty of Engineering, Thammasat University, Klong Luang, Pathumthani, 12121, Thailand
Abstract
This study proposes new electrocardiographic markers for prediction of ventricular arrhythmias. Two types of sinus rhythm were studied including sinus rhythm prior to ventricular arrhythmias (Pre-VF) group and normal sinus rhythm (SN) group. The technique based on wavelet transform was used for extracting electrographic markers. Three characteristic features of wavelet spectrum including frequency mean (FM), variance (VAR), and skewness (SK) were extracted from individual electrocardiogram (ECG) beat. We found that the wavelet spectrums in SN group were significantly broader than Pre-VF group. Furthermore the wavelet spectrums in Pre-VF group were more positively skewed distribution compared to SN group. Also, the FM marker achieved the best performance in separating the Pre-VF group from the SN group with the AUC of 98.9%, the sensitivity of 93%, and the specificity of 98%. Patients in the pre-VF group can be completely separated from healthy subjects using FM and SK as co-markers.
Publisher
National Taiwan University
Subject
Biomedical Engineering,Bioengineering,Biophysics