Automated artifact elimination of physiological signals using a deep belief network: An application for continuously measured arterial blood pressure waveforms

Yunsik Son, Seung Bo Lee, Hakseung Kim, Eun Suk Song, Hyub Huh, Marek Czosnyka, Dong Ju Kim

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Artifacts in physiological signals acquired during intensive care have the potential to be recognized as critical pathological events and lead to misdiagnosis or mismanagement. Manual artifact removal necessitates significant labor-time intensity and is subject to inter- and intra-observer variability. Various methods have been proposed to automate the task; however, the methods are yet to be validated, possibly due to the diversity of artifact types. Deep belief networks (DBNs) have been shown to be capable of learning generative and discriminative feature extraction models, hence suitable for classifying signals with multiple features. This study proposed a DBN-based model for artifact elimination in pulse waveform signals, which incorporates pulse segmentation, pressure normalization and decision models using DBN, and applied the model to artifact removal in monitoring arterial blood pressure (ABP). When compared with a widely used ABP artifact removal algorithm (signal abnormality index; SAI), the DBN model exhibited significantly higher classification performance (net prediction of optimal DBN = 95.9%, SAI = 84.7%). In particular, DBN exhibited greater sensitivity than SAI for identifying various types of artifacts (motion = 93.6%, biological = 95.4%, cuff inflation = 89.1%, transducer flushing = 97%). The proposed model could significantly enhance the quality of signal analysis, hence may be beneficial for use in continuous patient monitoring in clinical practice.

Original languageEnglish
Pages (from-to)145-158
Number of pages14
JournalInformation Sciences
Volume456
DOIs
Publication statusPublished - 2018 Aug 1

Fingerprint

Belief Networks
Blood pressure
Blood Pressure
Bayesian networks
Waveform
Elimination
Observer
Monitoring
Patient monitoring
Signal Analysis
Decision Model
Signal analysis
Transducer
Model
Inflation
Network Model
Feature Extraction
Normalization
Feature extraction
Transducers

Keywords

  • Arterial pressure
  • Artifacts
  • Computer-assisted
  • Monitoring
  • Neural networks (computer)
  • Physiologic
  • Signal processing

ASJC Scopus subject areas

  • Software
  • Control and Systems Engineering
  • Theoretical Computer Science
  • Computer Science Applications
  • Information Systems and Management
  • Artificial Intelligence

Cite this

Automated artifact elimination of physiological signals using a deep belief network : An application for continuously measured arterial blood pressure waveforms. / Son, Yunsik; Lee, Seung Bo; Kim, Hakseung; Song, Eun Suk; Huh, Hyub; Czosnyka, Marek; Kim, Dong Ju.

In: Information Sciences, Vol. 456, 01.08.2018, p. 145-158.

Research output: Contribution to journalArticle

Son, Yunsik ; Lee, Seung Bo ; Kim, Hakseung ; Song, Eun Suk ; Huh, Hyub ; Czosnyka, Marek ; Kim, Dong Ju. / Automated artifact elimination of physiological signals using a deep belief network : An application for continuously measured arterial blood pressure waveforms. In: Information Sciences. 2018 ; Vol. 456. pp. 145-158.
@article{3b0440709c0a415190c41316a7d37948,
title = "Automated artifact elimination of physiological signals using a deep belief network: An application for continuously measured arterial blood pressure waveforms",
abstract = "Artifacts in physiological signals acquired during intensive care have the potential to be recognized as critical pathological events and lead to misdiagnosis or mismanagement. Manual artifact removal necessitates significant labor-time intensity and is subject to inter- and intra-observer variability. Various methods have been proposed to automate the task; however, the methods are yet to be validated, possibly due to the diversity of artifact types. Deep belief networks (DBNs) have been shown to be capable of learning generative and discriminative feature extraction models, hence suitable for classifying signals with multiple features. This study proposed a DBN-based model for artifact elimination in pulse waveform signals, which incorporates pulse segmentation, pressure normalization and decision models using DBN, and applied the model to artifact removal in monitoring arterial blood pressure (ABP). When compared with a widely used ABP artifact removal algorithm (signal abnormality index; SAI), the DBN model exhibited significantly higher classification performance (net prediction of optimal DBN = 95.9{\%}, SAI = 84.7{\%}). In particular, DBN exhibited greater sensitivity than SAI for identifying various types of artifacts (motion = 93.6{\%}, biological = 95.4{\%}, cuff inflation = 89.1{\%}, transducer flushing = 97{\%}). The proposed model could significantly enhance the quality of signal analysis, hence may be beneficial for use in continuous patient monitoring in clinical practice.",
keywords = "Arterial pressure, Artifacts, Computer-assisted, Monitoring, Neural networks (computer), Physiologic, Signal processing",
author = "Yunsik Son and Lee, {Seung Bo} and Hakseung Kim and Song, {Eun Suk} and Hyub Huh and Marek Czosnyka and Kim, {Dong Ju}",
year = "2018",
month = "8",
day = "1",
doi = "10.1016/j.ins.2018.05.018",
language = "English",
volume = "456",
pages = "145--158",
journal = "Information Sciences",
issn = "0020-0255",
publisher = "Elsevier Inc.",

}

TY - JOUR

T1 - Automated artifact elimination of physiological signals using a deep belief network

T2 - An application for continuously measured arterial blood pressure waveforms

AU - Son, Yunsik

AU - Lee, Seung Bo

AU - Kim, Hakseung

AU - Song, Eun Suk

AU - Huh, Hyub

AU - Czosnyka, Marek

AU - Kim, Dong Ju

PY - 2018/8/1

Y1 - 2018/8/1

N2 - Artifacts in physiological signals acquired during intensive care have the potential to be recognized as critical pathological events and lead to misdiagnosis or mismanagement. Manual artifact removal necessitates significant labor-time intensity and is subject to inter- and intra-observer variability. Various methods have been proposed to automate the task; however, the methods are yet to be validated, possibly due to the diversity of artifact types. Deep belief networks (DBNs) have been shown to be capable of learning generative and discriminative feature extraction models, hence suitable for classifying signals with multiple features. This study proposed a DBN-based model for artifact elimination in pulse waveform signals, which incorporates pulse segmentation, pressure normalization and decision models using DBN, and applied the model to artifact removal in monitoring arterial blood pressure (ABP). When compared with a widely used ABP artifact removal algorithm (signal abnormality index; SAI), the DBN model exhibited significantly higher classification performance (net prediction of optimal DBN = 95.9%, SAI = 84.7%). In particular, DBN exhibited greater sensitivity than SAI for identifying various types of artifacts (motion = 93.6%, biological = 95.4%, cuff inflation = 89.1%, transducer flushing = 97%). The proposed model could significantly enhance the quality of signal analysis, hence may be beneficial for use in continuous patient monitoring in clinical practice.

AB - Artifacts in physiological signals acquired during intensive care have the potential to be recognized as critical pathological events and lead to misdiagnosis or mismanagement. Manual artifact removal necessitates significant labor-time intensity and is subject to inter- and intra-observer variability. Various methods have been proposed to automate the task; however, the methods are yet to be validated, possibly due to the diversity of artifact types. Deep belief networks (DBNs) have been shown to be capable of learning generative and discriminative feature extraction models, hence suitable for classifying signals with multiple features. This study proposed a DBN-based model for artifact elimination in pulse waveform signals, which incorporates pulse segmentation, pressure normalization and decision models using DBN, and applied the model to artifact removal in monitoring arterial blood pressure (ABP). When compared with a widely used ABP artifact removal algorithm (signal abnormality index; SAI), the DBN model exhibited significantly higher classification performance (net prediction of optimal DBN = 95.9%, SAI = 84.7%). In particular, DBN exhibited greater sensitivity than SAI for identifying various types of artifacts (motion = 93.6%, biological = 95.4%, cuff inflation = 89.1%, transducer flushing = 97%). The proposed model could significantly enhance the quality of signal analysis, hence may be beneficial for use in continuous patient monitoring in clinical practice.

KW - Arterial pressure

KW - Artifacts

KW - Computer-assisted

KW - Monitoring

KW - Neural networks (computer)

KW - Physiologic

KW - Signal processing

UR - http://www.scopus.com/inward/record.url?scp=85048833329&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85048833329&partnerID=8YFLogxK

U2 - 10.1016/j.ins.2018.05.018

DO - 10.1016/j.ins.2018.05.018

M3 - Article

AN - SCOPUS:85048833329

VL - 456

SP - 145

EP - 158

JO - Information Sciences

JF - Information Sciences

SN - 0020-0255

ER -