Empirical mode decomposition for chest compression and ventilation detection in cardiac arrest

Erik Alonso, Elisabete Aramendi, Digna González-Otero, Unai Ayala, Mohamud Ramzan Daya, James K. Russell

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The thoracic impedance (TI) signal, which reflects fluctuations due to CCs and ventilations, has been suggested as a surrogate to compute CC-rate and ventilation-rate during cardiopulmonary resuscitation. This study developed a method based on empirical mode decomposition (EMD) to compute CC-rate and ventilation-rate using exclusively the TI. Twenty out-of-hospital cardiac arrest episodes containing the TI, compression depth (gold standard for CC-rate), and capnography (gold standard for ventilation-rate) signals were used. The EMD decomposed the TI signal into intrinsic mode functions (IMFs). IMFs were combined based on their median instantaneous frequency to reconstruct separately the CC-signal and the ventilation-signal. Independent CC and ventilation detectors were used based on fixed thresholds for durations and dynamic thresholds for the amplitudes of the fluctuations. Sensitivity and positive predictive value (PPV) for each detector were 99.35%/98.75% and 93.21%/82.40%. CC-rate and ventilation-rate were computed based on instants of CCs and ventilations respectively. When comparing detected rates with rates obtained from the gold standards, the mean (SD) errors were 0.57(0.55) min-1 and 1.10 (1.19) min -1 for CC-rate and ventilation-rate respectively. We concluded that CC-rate and ventilation-rate can be accurately estimated applying EMD to the TI.

Original languageEnglish (US)
Title of host publicationComputing in Cardiology
PublisherIEEE Computer Society
Pages17-20
Number of pages4
Volume41
EditionJanuary
StatePublished - 2014
Event41st Computing in Cardiology Conference, CinC 2014 - Cambridge, United States
Duration: Sep 7 2014Sep 10 2014

Other

Other41st Computing in Cardiology Conference, CinC 2014
CountryUnited States
CityCambridge
Period9/7/149/10/14

Fingerprint

Heart Arrest
Ventilation
Thorax
Decomposition
Electric Impedance
Capnography
Resuscitation
Detectors
Out-of-Hospital Cardiac Arrest
Cardiopulmonary Resuscitation

ASJC Scopus subject areas

  • Cardiology and Cardiovascular Medicine
  • Computer Science(all)

Cite this

Alonso, E., Aramendi, E., González-Otero, D., Ayala, U., Daya, M. R., & Russell, J. K. (2014). Empirical mode decomposition for chest compression and ventilation detection in cardiac arrest. In Computing in Cardiology (January ed., Vol. 41, pp. 17-20). [7042968] IEEE Computer Society.

Empirical mode decomposition for chest compression and ventilation detection in cardiac arrest. / Alonso, Erik; Aramendi, Elisabete; González-Otero, Digna; Ayala, Unai; Daya, Mohamud Ramzan; Russell, James K.

Computing in Cardiology. Vol. 41 January. ed. IEEE Computer Society, 2014. p. 17-20 7042968.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Alonso, E, Aramendi, E, González-Otero, D, Ayala, U, Daya, MR & Russell, JK 2014, Empirical mode decomposition for chest compression and ventilation detection in cardiac arrest. in Computing in Cardiology. January edn, vol. 41, 7042968, IEEE Computer Society, pp. 17-20, 41st Computing in Cardiology Conference, CinC 2014, Cambridge, United States, 9/7/14.
Alonso E, Aramendi E, González-Otero D, Ayala U, Daya MR, Russell JK. Empirical mode decomposition for chest compression and ventilation detection in cardiac arrest. In Computing in Cardiology. January ed. Vol. 41. IEEE Computer Society. 2014. p. 17-20. 7042968
Alonso, Erik ; Aramendi, Elisabete ; González-Otero, Digna ; Ayala, Unai ; Daya, Mohamud Ramzan ; Russell, James K. / Empirical mode decomposition for chest compression and ventilation detection in cardiac arrest. Computing in Cardiology. Vol. 41 January. ed. IEEE Computer Society, 2014. pp. 17-20
@inproceedings{95b3e654d2694b1cbe9bd3b1f62f8b75,
title = "Empirical mode decomposition for chest compression and ventilation detection in cardiac arrest",
abstract = "The thoracic impedance (TI) signal, which reflects fluctuations due to CCs and ventilations, has been suggested as a surrogate to compute CC-rate and ventilation-rate during cardiopulmonary resuscitation. This study developed a method based on empirical mode decomposition (EMD) to compute CC-rate and ventilation-rate using exclusively the TI. Twenty out-of-hospital cardiac arrest episodes containing the TI, compression depth (gold standard for CC-rate), and capnography (gold standard for ventilation-rate) signals were used. The EMD decomposed the TI signal into intrinsic mode functions (IMFs). IMFs were combined based on their median instantaneous frequency to reconstruct separately the CC-signal and the ventilation-signal. Independent CC and ventilation detectors were used based on fixed thresholds for durations and dynamic thresholds for the amplitudes of the fluctuations. Sensitivity and positive predictive value (PPV) for each detector were 99.35{\%}/98.75{\%} and 93.21{\%}/82.40{\%}. CC-rate and ventilation-rate were computed based on instants of CCs and ventilations respectively. When comparing detected rates with rates obtained from the gold standards, the mean (SD) errors were 0.57(0.55) min-1 and 1.10 (1.19) min -1 for CC-rate and ventilation-rate respectively. We concluded that CC-rate and ventilation-rate can be accurately estimated applying EMD to the TI.",
author = "Erik Alonso and Elisabete Aramendi and Digna Gonz{\'a}lez-Otero and Unai Ayala and Daya, {Mohamud Ramzan} and Russell, {James K.}",
year = "2014",
language = "English (US)",
volume = "41",
pages = "17--20",
booktitle = "Computing in Cardiology",
publisher = "IEEE Computer Society",
edition = "January",

}

TY - GEN

T1 - Empirical mode decomposition for chest compression and ventilation detection in cardiac arrest

AU - Alonso, Erik

AU - Aramendi, Elisabete

AU - González-Otero, Digna

AU - Ayala, Unai

AU - Daya, Mohamud Ramzan

AU - Russell, James K.

PY - 2014

Y1 - 2014

N2 - The thoracic impedance (TI) signal, which reflects fluctuations due to CCs and ventilations, has been suggested as a surrogate to compute CC-rate and ventilation-rate during cardiopulmonary resuscitation. This study developed a method based on empirical mode decomposition (EMD) to compute CC-rate and ventilation-rate using exclusively the TI. Twenty out-of-hospital cardiac arrest episodes containing the TI, compression depth (gold standard for CC-rate), and capnography (gold standard for ventilation-rate) signals were used. The EMD decomposed the TI signal into intrinsic mode functions (IMFs). IMFs were combined based on their median instantaneous frequency to reconstruct separately the CC-signal and the ventilation-signal. Independent CC and ventilation detectors were used based on fixed thresholds for durations and dynamic thresholds for the amplitudes of the fluctuations. Sensitivity and positive predictive value (PPV) for each detector were 99.35%/98.75% and 93.21%/82.40%. CC-rate and ventilation-rate were computed based on instants of CCs and ventilations respectively. When comparing detected rates with rates obtained from the gold standards, the mean (SD) errors were 0.57(0.55) min-1 and 1.10 (1.19) min -1 for CC-rate and ventilation-rate respectively. We concluded that CC-rate and ventilation-rate can be accurately estimated applying EMD to the TI.

AB - The thoracic impedance (TI) signal, which reflects fluctuations due to CCs and ventilations, has been suggested as a surrogate to compute CC-rate and ventilation-rate during cardiopulmonary resuscitation. This study developed a method based on empirical mode decomposition (EMD) to compute CC-rate and ventilation-rate using exclusively the TI. Twenty out-of-hospital cardiac arrest episodes containing the TI, compression depth (gold standard for CC-rate), and capnography (gold standard for ventilation-rate) signals were used. The EMD decomposed the TI signal into intrinsic mode functions (IMFs). IMFs were combined based on their median instantaneous frequency to reconstruct separately the CC-signal and the ventilation-signal. Independent CC and ventilation detectors were used based on fixed thresholds for durations and dynamic thresholds for the amplitudes of the fluctuations. Sensitivity and positive predictive value (PPV) for each detector were 99.35%/98.75% and 93.21%/82.40%. CC-rate and ventilation-rate were computed based on instants of CCs and ventilations respectively. When comparing detected rates with rates obtained from the gold standards, the mean (SD) errors were 0.57(0.55) min-1 and 1.10 (1.19) min -1 for CC-rate and ventilation-rate respectively. We concluded that CC-rate and ventilation-rate can be accurately estimated applying EMD to the TI.

UR - http://www.scopus.com/inward/record.url?scp=84931384276&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84931384276&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84931384276

VL - 41

SP - 17

EP - 20

BT - Computing in Cardiology

PB - IEEE Computer Society

ER -