Tampere University of Technology

TUTCRIS Research Portal

Real-time Phonocardiogram Anomaly Detection by Adaptive 1D Convolu‐ tional Neural Networks

Research output: Contribution to journalArticleScientificpeer-review

Standard

Real-time Phonocardiogram Anomaly Detection by Adaptive 1D Convolu‐ tional Neural Networks. / Kiranyaz, Serkan; Zabihi, Morteza; Rad, Ali Bahrami; Ince, Turker; Hamila, Ridha; Gabbouj, Moncef.

In: Neurocomputing, 19.05.2020, p. 291-301.

Research output: Contribution to journalArticleScientificpeer-review

Harvard

APA

Vancouver

Author

Kiranyaz, Serkan ; Zabihi, Morteza ; Rad, Ali Bahrami ; Ince, Turker ; Hamila, Ridha ; Gabbouj, Moncef. / Real-time Phonocardiogram Anomaly Detection by Adaptive 1D Convolu‐ tional Neural Networks. In: Neurocomputing. 2020 ; pp. 291-301.

Bibtex - Download

@article{ccbd0aebe41d47cb865ea955a03f2f83,
title = "Real-time Phonocardiogram Anomaly Detection by Adaptive 1D Convolu‐ tional Neural Networks",
abstract = "The heart sound signals (Phonocardiogram - PCG) enable the earliest monitoring to detect a potential cardiovascular pathology and have recently become a crucial tool as a diagnostic test in outpatient monitoring to assess heart hemodynamic status. The need for an automated and accurate anomaly detection method for PCG has thus become imminent. To determine the state-of-the-art PCG classification algorithm, 48 international teams competed in the PhysioNet (CinC) Challenge in 2016 over the largest benchmark dataset with 3126 records with the classification outputs, normal (N), abnormal (A) and unsure – too noisy (U). In this study, our aim is to push this frontier further; however, we focus deliberately on the anomaly detection problem while assuming a reasonably high Signal-to-Noise Ratio (SNR) on the records. By using 1D Convolutional Neural Networks trained with a novel data purification approach, we aim to achieve the highest detection performance and real-time processing ability with significantly lower delay and computational complexity. The experimental results over the high-quality subset of the same benchmark dataset show that the proposed approach achieves both objectives. Furthermore, our findings reveal the fact that further improvements indeed require a personalized (patient-specific) approach to avoid major drawbacks of a global PCG classification approach.",
author = "Serkan Kiranyaz and Morteza Zabihi and Rad, {Ali Bahrami} and Turker Ince and Ridha Hamila and Moncef Gabbouj",
note = "EXT={"}Kiranyaz, Serkan{"}",
year = "2020",
month = "5",
day = "19",
doi = "10.1016/j.neucom.2020.05.063",
language = "English",
pages = "291--301",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier Science B.V.",

}

RIS (suitable for import to EndNote) - Download

TY - JOUR

T1 - Real-time Phonocardiogram Anomaly Detection by Adaptive 1D Convolu‐ tional Neural Networks

AU - Kiranyaz, Serkan

AU - Zabihi, Morteza

AU - Rad, Ali Bahrami

AU - Ince, Turker

AU - Hamila, Ridha

AU - Gabbouj, Moncef

N1 - EXT="Kiranyaz, Serkan"

PY - 2020/5/19

Y1 - 2020/5/19

N2 - The heart sound signals (Phonocardiogram - PCG) enable the earliest monitoring to detect a potential cardiovascular pathology and have recently become a crucial tool as a diagnostic test in outpatient monitoring to assess heart hemodynamic status. The need for an automated and accurate anomaly detection method for PCG has thus become imminent. To determine the state-of-the-art PCG classification algorithm, 48 international teams competed in the PhysioNet (CinC) Challenge in 2016 over the largest benchmark dataset with 3126 records with the classification outputs, normal (N), abnormal (A) and unsure – too noisy (U). In this study, our aim is to push this frontier further; however, we focus deliberately on the anomaly detection problem while assuming a reasonably high Signal-to-Noise Ratio (SNR) on the records. By using 1D Convolutional Neural Networks trained with a novel data purification approach, we aim to achieve the highest detection performance and real-time processing ability with significantly lower delay and computational complexity. The experimental results over the high-quality subset of the same benchmark dataset show that the proposed approach achieves both objectives. Furthermore, our findings reveal the fact that further improvements indeed require a personalized (patient-specific) approach to avoid major drawbacks of a global PCG classification approach.

AB - The heart sound signals (Phonocardiogram - PCG) enable the earliest monitoring to detect a potential cardiovascular pathology and have recently become a crucial tool as a diagnostic test in outpatient monitoring to assess heart hemodynamic status. The need for an automated and accurate anomaly detection method for PCG has thus become imminent. To determine the state-of-the-art PCG classification algorithm, 48 international teams competed in the PhysioNet (CinC) Challenge in 2016 over the largest benchmark dataset with 3126 records with the classification outputs, normal (N), abnormal (A) and unsure – too noisy (U). In this study, our aim is to push this frontier further; however, we focus deliberately on the anomaly detection problem while assuming a reasonably high Signal-to-Noise Ratio (SNR) on the records. By using 1D Convolutional Neural Networks trained with a novel data purification approach, we aim to achieve the highest detection performance and real-time processing ability with significantly lower delay and computational complexity. The experimental results over the high-quality subset of the same benchmark dataset show that the proposed approach achieves both objectives. Furthermore, our findings reveal the fact that further improvements indeed require a personalized (patient-specific) approach to avoid major drawbacks of a global PCG classification approach.

U2 - 10.1016/j.neucom.2020.05.063

DO - 10.1016/j.neucom.2020.05.063

M3 - Article

SP - 291

EP - 301

JO - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

ER -