Tampere University of Technology

TUTCRIS Research Portal

Acquiring respiration rate from photoplethysmographic signal by recursive bayesian tracking of intrinsic modes in time-frequency spectra

Research output: Contribution to journalArticleScientificpeer-review

Standard

Acquiring respiration rate from photoplethysmographic signal by recursive bayesian tracking of intrinsic modes in time-frequency spectra. / Pirhonen, Mikko; Peltokangas, Mikko; Vehkaoja, Antti.

In: Sensors, Vol. 18, No. 6, 1693, 01.06.2018.

Research output: Contribution to journalArticleScientificpeer-review

Harvard

APA

Vancouver

Author

Bibtex - Download

@article{d0f80a119c924caeb4aab5f0b1499e65,
title = "Acquiring respiration rate from photoplethysmographic signal by recursive bayesian tracking of intrinsic modes in time-frequency spectra",
abstract = "Respiration rate (RR) provides useful information for assessing the status of a patient. We propose RR estimation based on photoplethysmography (PPG) because the blood perfusion dynamics are known to carry information on breathing, as respiration-induced modulations in the PPG signal. We studied the use of amplitude variability of transmittance mode finger PPG signal in RR estimation by comparing four time-frequency (TF) representation methods of the signal cascaded with a particle filter. The TF methods compared were short-time Fourier transform (STFT) and three types of synchrosqueezing methods. The public VORTAL database was used in this study. The results indicate that the advanced frequency reallocation methods based on synchrosqueezing approach may present improvement over linear methods, such as STFT. The best results were achieved using wavelet synchrosqueezing transform, having a mean absolute error and median error of 2.33 and 1.15 breaths per minute, respectively. Synchrosqueezing methods were generally more accurate than STFT on most of the subjects when particle filtering was applied. While TF analysis combined with particle filtering is a promising alternative for real-time estimation of RR, artefacts and non-respiration-related frequency components remain problematic and impose requirements for further studies in the areas of signal processing algorithms an PPG instrumentation.",
keywords = "Particle filters, Photoplethysmography, Respiration, Synchrosqueezing, Time-frequency analysis",
author = "Mikko Pirhonen and Mikko Peltokangas and Antti Vehkaoja",
note = "int=TUT-BMT,{"}Pirhonen, Mikko{"}",
year = "2018",
month = "6",
day = "1",
doi = "10.3390/s18061693",
language = "English",
volume = "18",
journal = "Sensors",
issn = "1424-8220",
publisher = "MDPI",
number = "6",

}

RIS (suitable for import to EndNote) - Download

TY - JOUR

T1 - Acquiring respiration rate from photoplethysmographic signal by recursive bayesian tracking of intrinsic modes in time-frequency spectra

AU - Pirhonen, Mikko

AU - Peltokangas, Mikko

AU - Vehkaoja, Antti

N1 - int=TUT-BMT,"Pirhonen, Mikko"

PY - 2018/6/1

Y1 - 2018/6/1

N2 - Respiration rate (RR) provides useful information for assessing the status of a patient. We propose RR estimation based on photoplethysmography (PPG) because the blood perfusion dynamics are known to carry information on breathing, as respiration-induced modulations in the PPG signal. We studied the use of amplitude variability of transmittance mode finger PPG signal in RR estimation by comparing four time-frequency (TF) representation methods of the signal cascaded with a particle filter. The TF methods compared were short-time Fourier transform (STFT) and three types of synchrosqueezing methods. The public VORTAL database was used in this study. The results indicate that the advanced frequency reallocation methods based on synchrosqueezing approach may present improvement over linear methods, such as STFT. The best results were achieved using wavelet synchrosqueezing transform, having a mean absolute error and median error of 2.33 and 1.15 breaths per minute, respectively. Synchrosqueezing methods were generally more accurate than STFT on most of the subjects when particle filtering was applied. While TF analysis combined with particle filtering is a promising alternative for real-time estimation of RR, artefacts and non-respiration-related frequency components remain problematic and impose requirements for further studies in the areas of signal processing algorithms an PPG instrumentation.

AB - Respiration rate (RR) provides useful information for assessing the status of a patient. We propose RR estimation based on photoplethysmography (PPG) because the blood perfusion dynamics are known to carry information on breathing, as respiration-induced modulations in the PPG signal. We studied the use of amplitude variability of transmittance mode finger PPG signal in RR estimation by comparing four time-frequency (TF) representation methods of the signal cascaded with a particle filter. The TF methods compared were short-time Fourier transform (STFT) and three types of synchrosqueezing methods. The public VORTAL database was used in this study. The results indicate that the advanced frequency reallocation methods based on synchrosqueezing approach may present improvement over linear methods, such as STFT. The best results were achieved using wavelet synchrosqueezing transform, having a mean absolute error and median error of 2.33 and 1.15 breaths per minute, respectively. Synchrosqueezing methods were generally more accurate than STFT on most of the subjects when particle filtering was applied. While TF analysis combined with particle filtering is a promising alternative for real-time estimation of RR, artefacts and non-respiration-related frequency components remain problematic and impose requirements for further studies in the areas of signal processing algorithms an PPG instrumentation.

KW - Particle filters

KW - Photoplethysmography

KW - Respiration

KW - Synchrosqueezing

KW - Time-frequency analysis

U2 - 10.3390/s18061693

DO - 10.3390/s18061693

M3 - Article

VL - 18

JO - Sensors

JF - Sensors

SN - 1424-8220

IS - 6

M1 - 1693

ER -