Calculating Blood Pressure Based on Measured Heart Sounds

Lingguang Chen, Sean F. Wu, Yong Xu, William D. Lyman, Gaurav Kapur

Research output: Contribution to journalArticlepeer-review

3 Scopus citations

Abstract

The current standard technique for blood pressure determination is by using cuff/stethoscope, which is not suited for infants or children. Even for adults such an approach yields 60% accuracy with respect to intra-arterial blood pressure measurements. Moreover, it does not allow for continuous monitoring of blood pressure over 24 h and days. In this paper, a new methodology is developed that enables one to calculate the systolic and diastolic blood pressures continuously in a non-invasive manner based on the heart beats measured from the chest of a human being. To this end, we must separate the first and second heart sounds, known as S1 and S2, from the directly measured heart sound signals. Next, the individual characteristics of S1 and S2 must be identified and correlated to the systolic and diastolic blood pressures. It is emphasized that the material properties of a human being are highly inhomogeneous, changing from one organ to another, and the speed at which the heart sound signals propagate inside a human body cannot be determined precisely. Moreover, the exact locations from which the heart sounds are originated are unknown a priori, and must be estimated. As such, the computer model developed here is semi-empirical. Yet, validation results have demonstrated that this semi-empirical computer model can produce relatively robust and accurate calculations of the systolic and diastolic blood pressures with high statistical merits.

Original languageEnglish
Article number1750014
JournalJournal of Computational Acoustics
Volume25
Issue number3
DOIs
StatePublished - Sep 1 2017

Keywords

  • Wavelet denoising
  • blood pressure estimation
  • feature extraction
  • heart sound

Fingerprint

Dive into the research topics of 'Calculating Blood Pressure Based on Measured Heart Sounds'. Together they form a unique fingerprint.

Cite this