TY - GEN
T1 - An application of EMD technique in detection of tachycardia beats
AU - Desai, Usha
AU - Nayak, C. Gurudas
AU - Seshikala, G.
N1 - Publisher Copyright:
© 2016 IEEE.
PY - 2016/11/22
Y1 - 2016/11/22
N2 - An intelligent tachycardia diagnosis system assists the clinicians in discriminating normal and various tachycardia classes of heartbeats generally in the life-threatening conditions. This paper proposes, a methodology to classify multiclass tachycardia class using Electrocardiogram (ECG) signal. In this work, tachycardia classes are marked using nonlinear transform domain method Empirical Mode Decomposition (EMD). Using which tachycardia beats namely Atrial Flutter (AFL), Atrial Fibrillation (A-Fib), Ventricular Fibrillation (V-Fib) and Normal Sinus Rhythm (NSR) is discriminated. Independent Component Analysis (ICA) is applied on the patterns for dimensionality reduction and ten-fold cross validation is executed during the classifier development. Performance of diagnosis is compared individually using these three classifiers viz. Decision Tree (DT), Rotation Forest (ROF) and Random Forest (RAF) through Cohen's kappa statistic (κ), overall accuracy (%) and class specific accuracy (%). In current study, altogether 3858 ECG beats, belonging to four classes of tachycardia are used. The results obtained presents EMD coefficients clinical significance (p<0.0001). Besides, using RAF ensemble classifier we have achieved with an accuracy of 91.21% in diagnosis of tachycardia beats. The proposed approach can be used in the cardiac portable devices such as defibrillators and telemonitoring applications.
AB - An intelligent tachycardia diagnosis system assists the clinicians in discriminating normal and various tachycardia classes of heartbeats generally in the life-threatening conditions. This paper proposes, a methodology to classify multiclass tachycardia class using Electrocardiogram (ECG) signal. In this work, tachycardia classes are marked using nonlinear transform domain method Empirical Mode Decomposition (EMD). Using which tachycardia beats namely Atrial Flutter (AFL), Atrial Fibrillation (A-Fib), Ventricular Fibrillation (V-Fib) and Normal Sinus Rhythm (NSR) is discriminated. Independent Component Analysis (ICA) is applied on the patterns for dimensionality reduction and ten-fold cross validation is executed during the classifier development. Performance of diagnosis is compared individually using these three classifiers viz. Decision Tree (DT), Rotation Forest (ROF) and Random Forest (RAF) through Cohen's kappa statistic (κ), overall accuracy (%) and class specific accuracy (%). In current study, altogether 3858 ECG beats, belonging to four classes of tachycardia are used. The results obtained presents EMD coefficients clinical significance (p<0.0001). Besides, using RAF ensemble classifier we have achieved with an accuracy of 91.21% in diagnosis of tachycardia beats. The proposed approach can be used in the cardiac portable devices such as defibrillators and telemonitoring applications.
UR - http://www.scopus.com/inward/record.url?scp=85006741637&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85006741637&partnerID=8YFLogxK
U2 - 10.1109/ICCSP.2016.7754389
DO - 10.1109/ICCSP.2016.7754389
M3 - Conference contribution
AN - SCOPUS:85006741637
T3 - International Conference on Communication and Signal Processing, ICCSP 2016
SP - 1420
EP - 1424
BT - International Conference on Communication and Signal Processing, ICCSP 2016
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2016 International Conference on Communication and Signal Processing, ICCSP 2016
Y2 - 4 April 2016 through 6 April 2016
ER -