Classification of atrial fibrillation episodes from sparse electrocardiogram data

Satish Bukkapatnam, Ranga Komanduri, Hui Yang, Prahalad Rao, Wen Chen Lih, Milind Malshe, Lionel M. Raff, Bruce Benjamin, Mark Rockley

Research output: Contribution to journalArticle

14 Citations (Scopus)

Abstract

Background: Atrial fibrillation (AF) is the most common form of cardiac arrhythmia. This paper presents the application of the Classification and Regression Tree (CART) technique for detecting spontaneous termination or sustenance of AF with sparse data. Method: Electrocardiogram (ECG) recordings were obtained from the PhysioNet (AF Termination Challenge Database 2004) Web site. Signal analysis, feature extraction, and classification were made to distinguish among 3 AF episodes, namely, Nonterminating (N), Soon (<1 minute) to be terminating (S), and Terminating immediately (<1 second) (T). Results: A continuous wavelet transform whose basis functions match the EKG patterns was found to yield compact representation (∼2 orders of magnitude). This facilitates the development of efficient algorithms for beat detection, QRST subtraction, and multiple ECG quantifier extraction (eg, QRS width, QT interval). A compact feature set was extracted through principal component analysis of these quantifiers. Accuracies exceeding 90% for AF episode classification were achieved. Conclusions: A wavelet representation customized to the ECG signal pattern was found to yield 98% lower entropies compared with other representations that use standard library wavelets. The Classification and Regression Tree (CART) technique seems to distinguish the N vs T, and the S vs T classifications very accurately.

Original languageEnglish
Pages (from-to)292-299
Number of pages8
JournalJournal of Electrocardiology
Volume41
Issue number4
DOIs
StatePublished - 1 Jul 2008

Fingerprint

Atrial Fibrillation
Electrocardiography
Wavelet Analysis
Entropy
Principal Component Analysis
Libraries
Cardiac Arrhythmias
Databases

Keywords

  • Atrial fibrillation (AF)
  • CART decision tree
  • Electrocardiogram (ECG)
  • Feature extraction
  • Statistical analysis
  • Wavelet analysis

Cite this

Bukkapatnam, S., Komanduri, R., Yang, H., Rao, P., Lih, W. C., Malshe, M., ... Rockley, M. (2008). Classification of atrial fibrillation episodes from sparse electrocardiogram data. Journal of Electrocardiology, 41(4), 292-299. https://doi.org/10.1016/j.jelectrocard.2008.01.004
Bukkapatnam, Satish ; Komanduri, Ranga ; Yang, Hui ; Rao, Prahalad ; Lih, Wen Chen ; Malshe, Milind ; Raff, Lionel M. ; Benjamin, Bruce ; Rockley, Mark. / Classification of atrial fibrillation episodes from sparse electrocardiogram data. In: Journal of Electrocardiology. 2008 ; Vol. 41, No. 4. pp. 292-299.
@article{76a50d9fdaeb44baa9db17227c27664c,
title = "Classification of atrial fibrillation episodes from sparse electrocardiogram data",
abstract = "Background: Atrial fibrillation (AF) is the most common form of cardiac arrhythmia. This paper presents the application of the Classification and Regression Tree (CART) technique for detecting spontaneous termination or sustenance of AF with sparse data. Method: Electrocardiogram (ECG) recordings were obtained from the PhysioNet (AF Termination Challenge Database 2004) Web site. Signal analysis, feature extraction, and classification were made to distinguish among 3 AF episodes, namely, Nonterminating (N), Soon (<1 minute) to be terminating (S), and Terminating immediately (<1 second) (T). Results: A continuous wavelet transform whose basis functions match the EKG patterns was found to yield compact representation (∼2 orders of magnitude). This facilitates the development of efficient algorithms for beat detection, QRST subtraction, and multiple ECG quantifier extraction (eg, QRS width, QT interval). A compact feature set was extracted through principal component analysis of these quantifiers. Accuracies exceeding 90{\%} for AF episode classification were achieved. Conclusions: A wavelet representation customized to the ECG signal pattern was found to yield 98{\%} lower entropies compared with other representations that use standard library wavelets. The Classification and Regression Tree (CART) technique seems to distinguish the N vs T, and the S vs T classifications very accurately.",
keywords = "Atrial fibrillation (AF), CART decision tree, Electrocardiogram (ECG), Feature extraction, Statistical analysis, Wavelet analysis",
author = "Satish Bukkapatnam and Ranga Komanduri and Hui Yang and Prahalad Rao and Lih, {Wen Chen} and Milind Malshe and Raff, {Lionel M.} and Bruce Benjamin and Mark Rockley",
year = "2008",
month = "7",
day = "1",
doi = "10.1016/j.jelectrocard.2008.01.004",
language = "English",
volume = "41",
pages = "292--299",
journal = "Journal of Electrocardiology",
issn = "0022-0736",
publisher = "Churchill Livingstone",
number = "4",

}

Bukkapatnam, S, Komanduri, R, Yang, H, Rao, P, Lih, WC, Malshe, M, Raff, LM, Benjamin, B & Rockley, M 2008, 'Classification of atrial fibrillation episodes from sparse electrocardiogram data', Journal of Electrocardiology, vol. 41, no. 4, pp. 292-299. https://doi.org/10.1016/j.jelectrocard.2008.01.004

Classification of atrial fibrillation episodes from sparse electrocardiogram data. / Bukkapatnam, Satish; Komanduri, Ranga; Yang, Hui; Rao, Prahalad; Lih, Wen Chen; Malshe, Milind; Raff, Lionel M.; Benjamin, Bruce; Rockley, Mark.

In: Journal of Electrocardiology, Vol. 41, No. 4, 01.07.2008, p. 292-299.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Classification of atrial fibrillation episodes from sparse electrocardiogram data

AU - Bukkapatnam, Satish

AU - Komanduri, Ranga

AU - Yang, Hui

AU - Rao, Prahalad

AU - Lih, Wen Chen

AU - Malshe, Milind

AU - Raff, Lionel M.

AU - Benjamin, Bruce

AU - Rockley, Mark

PY - 2008/7/1

Y1 - 2008/7/1

N2 - Background: Atrial fibrillation (AF) is the most common form of cardiac arrhythmia. This paper presents the application of the Classification and Regression Tree (CART) technique for detecting spontaneous termination or sustenance of AF with sparse data. Method: Electrocardiogram (ECG) recordings were obtained from the PhysioNet (AF Termination Challenge Database 2004) Web site. Signal analysis, feature extraction, and classification were made to distinguish among 3 AF episodes, namely, Nonterminating (N), Soon (<1 minute) to be terminating (S), and Terminating immediately (<1 second) (T). Results: A continuous wavelet transform whose basis functions match the EKG patterns was found to yield compact representation (∼2 orders of magnitude). This facilitates the development of efficient algorithms for beat detection, QRST subtraction, and multiple ECG quantifier extraction (eg, QRS width, QT interval). A compact feature set was extracted through principal component analysis of these quantifiers. Accuracies exceeding 90% for AF episode classification were achieved. Conclusions: A wavelet representation customized to the ECG signal pattern was found to yield 98% lower entropies compared with other representations that use standard library wavelets. The Classification and Regression Tree (CART) technique seems to distinguish the N vs T, and the S vs T classifications very accurately.

AB - Background: Atrial fibrillation (AF) is the most common form of cardiac arrhythmia. This paper presents the application of the Classification and Regression Tree (CART) technique for detecting spontaneous termination or sustenance of AF with sparse data. Method: Electrocardiogram (ECG) recordings were obtained from the PhysioNet (AF Termination Challenge Database 2004) Web site. Signal analysis, feature extraction, and classification were made to distinguish among 3 AF episodes, namely, Nonterminating (N), Soon (<1 minute) to be terminating (S), and Terminating immediately (<1 second) (T). Results: A continuous wavelet transform whose basis functions match the EKG patterns was found to yield compact representation (∼2 orders of magnitude). This facilitates the development of efficient algorithms for beat detection, QRST subtraction, and multiple ECG quantifier extraction (eg, QRS width, QT interval). A compact feature set was extracted through principal component analysis of these quantifiers. Accuracies exceeding 90% for AF episode classification were achieved. Conclusions: A wavelet representation customized to the ECG signal pattern was found to yield 98% lower entropies compared with other representations that use standard library wavelets. The Classification and Regression Tree (CART) technique seems to distinguish the N vs T, and the S vs T classifications very accurately.

KW - Atrial fibrillation (AF)

KW - CART decision tree

KW - Electrocardiogram (ECG)

KW - Feature extraction

KW - Statistical analysis

KW - Wavelet analysis

UR - http://www.scopus.com/inward/record.url?scp=45049085326&partnerID=8YFLogxK

U2 - 10.1016/j.jelectrocard.2008.01.004

DO - 10.1016/j.jelectrocard.2008.01.004

M3 - Article

C2 - 18367198

AN - SCOPUS:45049085326

VL - 41

SP - 292

EP - 299

JO - Journal of Electrocardiology

JF - Journal of Electrocardiology

SN - 0022-0736

IS - 4

ER -