Learning Optimal Time-Frequency Representations for Heart Sound: A Comparative Study

Zhihua Wang; Zhihao Bao; Kun Qian; Bin Hu; Björn W. Schuller; Yoshiharu Yamamoto

doi:10.1007/978-981-19-4703-2_8

Learning Optimal Time-Frequency Representations for Heart Sound: A Comparative Study

Zhihua Wang
, Zhihao Bao
, Kun Qian
, Bin Hu
, Björn W. Schuller
, Yoshiharu Yamamoto

Beijing Institute of Technology
China University of Mining and Technology
University of Tokyo
Imperial College London

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

Computer audition based methods have increasingly attracted efforts among the community of digital health. In particular, heart sound analysis can provide a non-invasive, real-time, and convenient (anywhere and anytime) solution for preliminary diagnosis and/or long-term monitoring of patients who are suffering from cardiovascular diseases. Nevertheless, extracting excellent time-frequency features from the heart sound is not an easy task. On the one hand, heart sound belongs to audio signals, which may be suitable to be analysed by classic audio/speech techniques. On the other hand, this kind of sound generated by our human body should contain some characteristics of physiological signals. To this end, we propose a comprehensive investigation on time-frequency methods for analysing the heart sound, i.e., short-time Fourier transformation, wavelet transformation, Hilbert-Huang transformation, and Log-Mel transformation. The time-frequency representations will be automatically learnt via pre-trained deep convolutional neural networks. Experimental results show that all the investigated methods can reach a mean accuracy higher than 60.0%. Moreover, we find that wavelet transformation can beat other methods by reaching the highest mean accuracy of 75.1% in recognising normal or abnormal heart sounds.

Original language	English
Title of host publication	Proceedings of the 9th Conference on Sound and Music Technology - Revised Selected Papers from CMST
Editors	Xi Shao, Kun Qian, Xin Wang, Kejun Zhang
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	93-104
Number of pages	12
ISBN (Print)	9789811947025
DOIs	https://doi.org/10.1007/978-981-19-4703-2_8
State	Published - 2023
Externally published	Yes
Event	9th Conference on Sound and Music Technology, CSMT 2021 - Virtual, Online Duration: 1 Jun 2022 → …

Publication series

Name	Lecture Notes in Electrical Engineering
Volume	923
ISSN (Print)	1876-1100
ISSN (Electronic)	1876-1119

Conference

Conference	9th Conference on Sound and Music Technology, CSMT 2021
City	Virtual, Online
Period	1/06/22 → …

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

Computer audition
Deep learning
Digital health
Heart sound
Time-frequency analysis
Transfer learning

Access to Document

10.1007/978-981-19-4703-2_8

Cite this

Wang, Z., Bao, Z., Qian, K., Hu, B., Schuller, B. W., & Yamamoto, Y. (2023). Learning Optimal Time-Frequency Representations for Heart Sound: A Comparative Study. In X. Shao, K. Qian, X. Wang, & K. Zhang (Eds.), Proceedings of the 9th Conference on Sound and Music Technology - Revised Selected Papers from CMST (pp. 93-104). (Lecture Notes in Electrical Engineering; Vol. 923). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-19-4703-2_8

Wang, Zhihua ; Bao, Zhihao ; Qian, Kun et al. / Learning Optimal Time-Frequency Representations for Heart Sound : A Comparative Study. Proceedings of the 9th Conference on Sound and Music Technology - Revised Selected Papers from CMST. editor / Xi Shao ; Kun Qian ; Xin Wang ; Kejun Zhang. Springer Science and Business Media Deutschland GmbH, 2023. pp. 93-104 (Lecture Notes in Electrical Engineering).

@inproceedings{3fab0ba980fb4f419239384b693871df,

title = "Learning Optimal Time-Frequency Representations for Heart Sound: A Comparative Study",

abstract = "Computer audition based methods have increasingly attracted efforts among the community of digital health. In particular, heart sound analysis can provide a non-invasive, real-time, and convenient (anywhere and anytime) solution for preliminary diagnosis and/or long-term monitoring of patients who are suffering from cardiovascular diseases. Nevertheless, extracting excellent time-frequency features from the heart sound is not an easy task. On the one hand, heart sound belongs to audio signals, which may be suitable to be analysed by classic audio/speech techniques. On the other hand, this kind of sound generated by our human body should contain some characteristics of physiological signals. To this end, we propose a comprehensive investigation on time-frequency methods for analysing the heart sound, i.e., short-time Fourier transformation, wavelet transformation, Hilbert-Huang transformation, and Log-Mel transformation. The time-frequency representations will be automatically learnt via pre-trained deep convolutional neural networks. Experimental results show that all the investigated methods can reach a mean accuracy higher than 60.0\%. Moreover, we find that wavelet transformation can beat other methods by reaching the highest mean accuracy of 75.1\% in recognising normal or abnormal heart sounds.",

keywords = "Computer audition, Deep learning, Digital health, Heart sound, Time-frequency analysis, Transfer learning",

author = "Zhihua Wang and Zhihao Bao and Kun Qian and Bin Hu and Schuller, \{Bj{\"o}rn W.\} and Yoshiharu Yamamoto",

note = "Publisher Copyright: {\textcopyright} 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.; 9th Conference on Sound and Music Technology, CSMT 2021 ; Conference date: 01-06-2022",

year = "2023",

doi = "10.1007/978-981-19-4703-2\_8",

language = "English",

isbn = "9789811947025",

series = "Lecture Notes in Electrical Engineering",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "93--104",

editor = "Xi Shao and Kun Qian and Xin Wang and Kejun Zhang",

booktitle = "Proceedings of the 9th Conference on Sound and Music Technology - Revised Selected Papers from CMST",

}

Wang, Z, Bao, Z, Qian, K, Hu, B, Schuller, BW & Yamamoto, Y 2023, Learning Optimal Time-Frequency Representations for Heart Sound: A Comparative Study. in X Shao, K Qian, X Wang & K Zhang (eds), Proceedings of the 9th Conference on Sound and Music Technology - Revised Selected Papers from CMST. Lecture Notes in Electrical Engineering, vol. 923, Springer Science and Business Media Deutschland GmbH, pp. 93-104, 9th Conference on Sound and Music Technology, CSMT 2021, Virtual, Online, 1/06/22. https://doi.org/10.1007/978-981-19-4703-2_8

Learning Optimal Time-Frequency Representations for Heart Sound: A Comparative Study. / Wang, Zhihua; Bao, Zhihao; Qian, Kun et al.
Proceedings of the 9th Conference on Sound and Music Technology - Revised Selected Papers from CMST. ed. / Xi Shao; Kun Qian; Xin Wang; Kejun Zhang. Springer Science and Business Media Deutschland GmbH, 2023. p. 93-104 (Lecture Notes in Electrical Engineering; Vol. 923).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Learning Optimal Time-Frequency Representations for Heart Sound

T2 - 9th Conference on Sound and Music Technology, CSMT 2021

AU - Wang, Zhihua

AU - Bao, Zhihao

AU - Qian, Kun

AU - Hu, Bin

AU - Schuller, Björn W.

AU - Yamamoto, Yoshiharu

PY - 2023

Y1 - 2023

N2 - Computer audition based methods have increasingly attracted efforts among the community of digital health. In particular, heart sound analysis can provide a non-invasive, real-time, and convenient (anywhere and anytime) solution for preliminary diagnosis and/or long-term monitoring of patients who are suffering from cardiovascular diseases. Nevertheless, extracting excellent time-frequency features from the heart sound is not an easy task. On the one hand, heart sound belongs to audio signals, which may be suitable to be analysed by classic audio/speech techniques. On the other hand, this kind of sound generated by our human body should contain some characteristics of physiological signals. To this end, we propose a comprehensive investigation on time-frequency methods for analysing the heart sound, i.e., short-time Fourier transformation, wavelet transformation, Hilbert-Huang transformation, and Log-Mel transformation. The time-frequency representations will be automatically learnt via pre-trained deep convolutional neural networks. Experimental results show that all the investigated methods can reach a mean accuracy higher than 60.0%. Moreover, we find that wavelet transformation can beat other methods by reaching the highest mean accuracy of 75.1% in recognising normal or abnormal heart sounds.

AB - Computer audition based methods have increasingly attracted efforts among the community of digital health. In particular, heart sound analysis can provide a non-invasive, real-time, and convenient (anywhere and anytime) solution for preliminary diagnosis and/or long-term monitoring of patients who are suffering from cardiovascular diseases. Nevertheless, extracting excellent time-frequency features from the heart sound is not an easy task. On the one hand, heart sound belongs to audio signals, which may be suitable to be analysed by classic audio/speech techniques. On the other hand, this kind of sound generated by our human body should contain some characteristics of physiological signals. To this end, we propose a comprehensive investigation on time-frequency methods for analysing the heart sound, i.e., short-time Fourier transformation, wavelet transformation, Hilbert-Huang transformation, and Log-Mel transformation. The time-frequency representations will be automatically learnt via pre-trained deep convolutional neural networks. Experimental results show that all the investigated methods can reach a mean accuracy higher than 60.0%. Moreover, we find that wavelet transformation can beat other methods by reaching the highest mean accuracy of 75.1% in recognising normal or abnormal heart sounds.

KW - Computer audition

KW - Deep learning

KW - Digital health

KW - Heart sound

KW - Time-frequency analysis

KW - Transfer learning

UR - https://www.scopus.com/pages/publications/85137987305

U2 - 10.1007/978-981-19-4703-2_8

DO - 10.1007/978-981-19-4703-2_8

M3 - Conference contribution

AN - SCOPUS:85137987305

SN - 9789811947025

T3 - Lecture Notes in Electrical Engineering

SP - 93

EP - 104

BT - Proceedings of the 9th Conference on Sound and Music Technology - Revised Selected Papers from CMST

A2 - Shao, Xi

A2 - Qian, Kun

A2 - Wang, Xin

A2 - Zhang, Kejun

PB - Springer Science and Business Media Deutschland GmbH

Y2 - 1 June 2022

ER -

Wang Z, Bao Z, Qian K, Hu B, Schuller BW, Yamamoto Y. Learning Optimal Time-Frequency Representations for Heart Sound: A Comparative Study. In Shao X, Qian K, Wang X, Zhang K, editors, Proceedings of the 9th Conference on Sound and Music Technology - Revised Selected Papers from CMST. Springer Science and Business Media Deutschland GmbH. 2023. p. 93-104. (Lecture Notes in Electrical Engineering). doi: 10.1007/978-981-19-4703-2_8