Beat-to-beat cardiovascular hemodynamic parameters based on wavelet spectrogram of impedance data

https://doi.org/10.1016/j.bspc.2017.03.012Get rights and content

Highlights

  • Impedance cardiography is attractive low cost express-diagnostics.

  • We use wavelet analysis to improve and infuse new life into impedance cardiography.

  • Wavelets is useful for noise filtration and interpretation of the impedance cardiogram.

  • Wavelet processing recognises artifacts, breath wave, and morphological peculiarities.

Abstract

Impedance cardiography (ICG) is an attractive non-invasive tool for determining cardiac output, systemic vascular resistance and stroke volume (SV). New methods of improving the signal to noise ratio and data processing can infuse new life into this technique. In the present paper, wavelet transform is used to find solutions to two interrelated problems: (1) filtering of tetrapolar thoracic rheography signals (ICG, electrocardiogram and phonocardiogram (PCG)) to suppress random noise and respiratory variation; (2) quantitative description of the ICG and PCG parameters based on the analysis of two-dimensional time-frequency distributions of wavelet coefficients. The wavelet images provide an illustrative representation of the results obtained and allow us to define the ejection time and to control beat-to-beat changes in cardiac parameters in the systolic and diastolic phases of the cardiac cycle. The advantage of the wavelet-based approach is demonstrated by an example of determining the SV variation during the respiratory cycle. Wavelet-based processing of impedance cardiography signals reduces the influence of artifacts and the interference of breathing modulation, and displays the morphological features of cardiac cycles.

Introduction

In cardiovascular therapy, impedance cardiography (ICG) has long been recognised as an effective non-invasive method of determining cardiac output and systemic vascular resistance. It enables researchers to obtain one of the key hemodynamic parameters, the stroke volume (SV). The principle underlying the rheography method, according to which changes in the electrical resistance of a particular domain of the body to high-frequency alternating current are proportional to variations in blood volume in this domain, has been known for over 50 years. H. Mann published the results of his early impedancemetry studies of the human body in 1937. Kedrov first proposed a method to determine SV using bipolar body rheography in 1948 [1]; Nyboer et al. independently did so in 1950 [2]. According to Nyboer's study, the area under the rheogram curve can be related to the amount of blood ejected into the systole, i.e., the SV.

The tetrapolar thoracic rheography method, developed for NASA by a team of researchers under the guidance of Kubicek, was introduced in medical practice in the 1970s. Following this method four electrodes (two current-carrying and two potential probe electrodes) are placed on the neck and chest. This research led to the development of the Minnesota Impedance Cardiograph and a new formula [3], which relates the SV to the maximum value of the first derivative of the impedance waveform and the Left Ventricular Ejection Time (LVET) asSV=CLVETdZdtmax,where C is a parameter which depends on blood resistivity, distance between receiving electrodes and basic impedance of the body segment limited by receiving electrodes.

In a rather short period of time, the tetrapolar thoracic ICG was widely applied in practical medicine and highly rated by cardiologists in the 1970s and 1980s. However, during this time researchers (and especially practicing physicians) lost interest in this technique for a number of objective reasons, including the imperfection of measurement devices used for impedancemetry and a lack of digital techniques for processing ICG signals, which complicated the interpretation of the obtained results. Nevertheless, impedance cardiography techniques, in particular the tetrapolar thoracic ICG method, were still rather attractive to specialists due to their simplicity, low cost and probability of obtaining sufficiently accurate and reproducible express-results [4], [5], [6], [7], [8]. The application of advanced equipment and new methods of data processing can infuse new life into this technique.

Under ideal condition, the recorded ICG signals should reflect only physiological processes and allow one to determine the parameters of the permanent and variable component of the impedance. In practice, bioimpedance measurements are very sensitive to external electromagnetic effects and different risk factors (loose electrodes, patient movement, etc.). Such disturbances (artifacts) introduce large errors into hemodynamic parameters, and therefore real physiological processes can be lost among these artifacts.

Recently various denoising techniques were applied for suppressing the respiratory and motion artifacts in ICG, e.g., discrete wavelets, LMS-based adaptive filtering technique and EEMD denoising [9], [10], [11]. Denoising of signals with real respiratory and motion artifacts indicated effective suppression of artifacts without any significant signal distortion. Note, that the respiratory wave is mainly considered as an artifact, which should be excluded for valid estimation of SV. However, the SV modulation during different respiratory phases can provide important information about the function of the heart [12]. Wavelet spectrograms allow us not only to separate the respiratory and pulse waves, but also to analyse the variation of pulse wave parameters with the phase and depth of breathing. This is the main difference between the approach employed in the present study and the existing techniques.

In some cases, random errors in estimation of the sample mean can be reduced by increasing the size of the sample, but the contribution of the noise and artifacts persist in statistical characteristics such as variance or correlation. However, the variability of the hemodynamic parameters (e.g., the respiratory variation in SV) opens considerable scope for diagnosis of pathologies. Studying the variability, one should distinguish between the regular variability, which actually affects each cycle (due to the respiration or some functional tests), and the variability, which occurs during extraordinary and rare cycles that differ significantly from most cardiac cycles. The latter changes can be used to diagnose heart rhythm disturbances (for example, ventricular extrasystole) and chronic cardiac insufficiency [13]. Therefore, we aim to develop robust computational techniques for calculating a hemodynamic parameter for each tested cardiac cycle of ICG signals.

Among the mathematical techniques developed last decades for analysis of nonstationary multiscale processes, wavelets are a promising tool, particularly for analysis of cardio-related signals [14], [15], [16], [17], [18]. Unlike the Fourier transform, in which the basis functions cover the entire signal range, the wavelet analysis deals with the local time-scale features of the signal. We have carried out preliminary signal processing in order to find the self similar cardiac cycles. Then the wavelet transform of the signal has been used to solve two problems. The first one is a standard wavelet filtration of the tetrapolar thoracic rheography signals (impedance cardiogram, electrocardiogram and phonocardiogram) to suppress random noise and respiratory variation. The second problem concerns the determination of quantitative parameters of impedance cardiograms and is solved by analysing the two-dimensional time-frequency distributions of wavelet coefficients. Wavelet images provide an illustrative representation of the obtained results and allow us to control beat-to-beat changes of cardiac parameters in the systolic and diastolic phase of the cardiac cycle.

Section snippets

Wavelets

The signals that characterise cardiovascular activity (electrocadiograms, phonograms, impedance cardiograms, etc.) have an implicit quasi-periodic character. Note that analysis of the state of the cardiovascular system involves evaluating both the frequency variability of the main cardiac cycle and the structure variability of the signal inside a single cycle. It is also interesting that the characteristics obtained for patients at rest and those obtained during different functional tests are

Measurements and data preprocessing

Changes in the hemodynamic state of patients are registered and analysed using the thoracic tetrapolar polyrheocardiography method. We used a polygraphic lab-made measuring system based on the synchronous online monitoring ICG, electrocardiogram (ECG) and phonocardiogram (PCG) [17]. The measuring system includes the ICG recording channel at 100 kHz with an amplifier and demodulation, the standard two-electrode ECG channel and the PCG channel equipped by a low-frequency amplifier. In order to

Wavelet indices of ICG

The wavelet spectrogram of the impedance cardiogram is constructed using the sequence of cycles obtained after filtration, wavelet differentiation and cycle selection based on the similarity criterion. On the wavelet spectrogram of the impedance cardiogram (Fig. 5b) one can observe a sequence of localised structures of different scales (red-coloured region correspond to positive and blue-coloured region to negative variations). Each cardiac cycle has two clear-cut maxima.

In the description of

Study limitations

Special emphasis should be placed on the limitations of our investigations. First, arrhythmia or respiratory difficulties can change the breathing waveform so that higher harmonics will then contribute significantly to the amplitude of the wavelet coefficients at the scale of the E-wave. Also, with rapid breathing, a separation of the respiratory and cardiac components is not possible. Second, our results were obtained for healthy young adults and more research is needed to identify the details

Conclusions

The proposed approach presents a new method of processing impedance cardiography data based on wavelet spectrograms. After preliminary signal processing the wavelet plane of the impedance cardiogram is constructed. The values of the local maximum of the wavelet coefficients define the amplitudes of E-waves related to the systolic phase and O-waves related to the diastolic phase of the cardiac cycle. In some cases the wavelet spectrogram clearly display the morphological peculiarities of the

Acknowledgment

The work is supported by the Russian Science Foundation under project 14-15-00809.

References (31)

  • J. Van De Water et al.

    Impedance cardiography the next vital sign technology?

    Chest

    (2003)
  • B. Pickett et al.

    Usefulness of the impedance cardiogram to reflect left-ventricular diastolic function

    Am. J. Cardiol.

    (1993)
  • A. Kedrov

    An attempt to quantify the central and peripheral circulation by electrometric

    Clin. Med.

    (1948)
  • J. Nyboer et al.

    Radiocardiograms: electrical impedance changes of the heart in relation to electrocardiograms and heart sounds

    J. Clin. Invest.

    (1950)
  • W. Kubicek et al.

    Development and evaluation of an impedance cardiac output system

    Aerosp. Med.

    (1966)
  • G. Cybulski et al.

    Stroke volume and systolic time intervals: beat-to-beat comparison between echocardiography and ambulatory impedance cardiography in supine and tilted positions

    Med. Biol. Eng. Comput.

    (2004)
  • M. Nakagawara et al.

    A portable instrument for non-invasive monitoring of beat-by-beat cardiovascular haemodynamic parameters based on the volume-compensation and electrical-admittance method

    Med. Biol. Eng. Comput.

    (2000)
  • S. Taler et al.

    Resistant hypertension – comparing hemodynamic management to specialist care

    Hypertension

    (2002)
  • M. Van Eijnatten et al.

    Comparison of cardiac time intervals between echocardiography and impedance cardiography at various heart rates

    J. Electr. Bioimpedance

    (2014)
  • T. Sebastian et al.

    Wavelet based denoising for suppression of respiratory and motion artifacts in impedance cardiography

    Comput. Cardiol.

    (2011)
  • V. Pandey et al.

    Adaptive filtering for suppression of respiratory artifact in impedance cardiography

  • S. De Ridder et al.

    Comparison between EEMD, wavelet and fir denoising: influence on event detection in impedance cardiography

  • J.E. Hall et al.

    Textbook of Medical Physiology

    (2006)
  • M. Zubarev et al.

    Beat-to-beat variability of stroke volume output velocity measured by an impedance cardiographic method

    IFMBE Proc.

    (2007)
  • S. Chen

    A wavelet-based heart rate variability analysis for the study of nonsustained ventricular tachycardia

    IEEE Trans. Biomed. Eng.

    (2002)
  • Cited by (0)

    View full text