A modelling framework for assessment of arterial compliance by fusion of oscillometry and pulse wave velocity information

https://doi.org/10.1016/j.cmpb.2020.105492Get rights and content

Highlights

  • Novel approach for measuring arterial stiffness by signal fusion.

  • Bayesian principles applied to hemodynamic parameter measurement.

  • Easy-to-use, robust arterial stiffness measurement using standard technology.

  • Simulation framework for arterial stiffness measurement and algorithm development.

  • Method for data fusion of ECG, PPG and cuff-based blood pressure measurement.

Abstract

Background and Objectives

Measurement of arterial compliance is recognized as important for clinical use and for enabling better understanding of circulatory system regulation mechanisms. Estimation of arterial compliance involves either a direct measure of the ratio between arterial volume and pressure changes or an inference from the pulse wave velocity (PWV). In this study we demonstrate an approach to assess arterial compliance by fusion of these two information sources. The approach is based on combining oscillometry as used for blood pressure inference and PWV measurements based on ECG/PPG. Enabling reliable arterial compliance measurements will contribute to the understanding of regulation mechanisms of the arterial tree, possibly establishing arterial compliance as a key measure relevant in hemodynamic monitoring.

Methods

A measurement strategy, a physiological model, and a framework based on Bayesian principles are developed for measuring changes in arterial compliance based on combining oscillometry and PWV data. A simulation framework is used to study and validate the algorithm and measurement principle in detail, motivated by previous experimental findings.

Results

Simulations demonstrate the possibility of inferring arterial compliance via fusion of simultaneously acquired volume/pressure relationships and PWV data. In addition, the simulation framework demonstrates how Bayesian principles can be used to handle low signal – to – noise ratio and partial information loss.

Conclusions

The developed simulation framework shows the feasibility of the proposed approach for assessment of arterial compliance by combining multiple data sources. This represents a first step towards integration of arterial compliance measurements in hemodynamic monitoring using existing clinical technology. The Bayesian approach is of particular relevance for such patient monitoring settings, where measurements are repeated frequently, context is relevant, and data is affected by artefacts. In addition, the simulation framework is necessary for future clinical-study design, in order to determine device specifications and the extent to which noise affects the inference process.

Introduction

Up until now, research on measurement of vascular wall properties has mainly been conducted with the aim of identifying long-term impact of lifestyle, ageing or pathological conditions on cardiovascular function. In particular, studies related to long-term changes of arterial wall stiffness have contributed to better understanding of cardiovascular health and risks associated with cardiovascular disease. As a result, recent international efforts have been made to achieve standardization of arterial stiffness measurements as complementary to blood pressure measurements [1], [2], [3].

However, short-term, dynamic regulation of arteries is also relevant, especially in the context of hemodynamic monitoring. Wall stress and lumen diameter, adjustments of the smooth muscle tone are important for regulation of blood pressure (via the balance of stressed and unstressed volume and shifts of blood volume within the circulatory system [4]). Changes in these vascular mechanical properties can precede changes in blood pressure and cardiac output; knowledge of arterial compliance could be of high clinical interest. Critical care, trauma, ICU clinicians have expressed interest in these parameters [8], however, measurements are scarce. Methods for measuring and interpreting short-term changes in arterial compliance have yet to be investigated and developed.

Changes in smooth muscle tone have been identified as particularly relevant in hemodynamic measurements requiring calibration. Devices which interpret pulse wave velocity as surrogate for blood pressure changes [5,6], or devices which use pulse waveform as surrogate of cardiac output (PiCCO [7]), require re-calibration whenever changes in arterial compliance occur as result of hemodynamic regulation mechanisms, vasoactive drug administration or in cases of development of vascular edema [8]. Therefore, identifying when such devices require re-calibration can be a first relevant application to introduce arterial compliance measurements in clinical hemodynamic monitoring practice.

Such application can serve as basis for developing better understanding of smooth tone regulation and tackling other unmet clinical needs:

  • Earlier warning of patient deterioration; changes in arterial properties might precede changes in blood pressure, flow and may be used for early warning of hypotension and hemodynamic instability.

  • The uncertainty in interpretation of blood pressure (BP) values in hemodynamic monitoring [9], [10], [11], [12]; complementing the BP value with more information could make the measurement more specific (e.g. identify the cause of a BP drop – differentiate between heart related factors or vasodilation).

  • The dissociation between micro and macro circulation [13, 14]; reliable measurement of arterial compliance might give insights into regulation mechanisms at micro vs macro level.

  • The recognized need of including functional variables [15]; the cuff inflation could induce short-term changes in arterial properties, which might reveal information about the hemodynamic status of a patient [32].

  • Guidance of fluid management through addition of extra hemodynamic parameters related to arterial properties.

  • Research on oscillometric non-invasive BP measurements is needed to improve accuracy of the measurement especially for hypo- and hyper tension [31]. Understanding of arterial property expression in the oscillometric signal can improve accuracy of the BP measurement [19].

There are two existing methods of measuring arterial stiffness: either based on analysis of arterial cross-sectional area with respect to arterial pulsatile pressure or based on analysis of wave propagation along regions of the arterial tree [16]. Below we describe how these two methods can be implemented.

1. A suitable modality for measuring arterial cross-sectional area is via imaging techniques such as ultrasound [17] or MRI [18, 28]. However, such techniques are not practical in acute clinical care settings. Alternatively, the arterial size can be obtained more indirectly via oscillometry. Oscillometric measurements are based on varying transmural pressure across the brachial arterial wall by inflation of a pneumatic cuff. A pressure sensor inside the cuff gives indication on the amplitudes of the arterial volume pulses via the resulting cuff pressure pulses, as transmural pressure over the artery is varied.

The data obtained from oscillometry is usually processed to derive blood pressure values. Mean arterial pressure is found by identifying the cuff pressure at which maximum volume pulses occur, while systolic and diastolic pressures are found at cuff pressures where certain empirical ratios in the pressure signal occur [33]. For example, cuff pressure on the falling phase of the oscillation signal at which amplitude of pulsation is 50% of the maximum amplitude is found to coincide with systolic BP. Cuff pressure on the rising phase of the oscillation signal at which amplitude of pulsation is 70% of the maximum amplitude is found to coincide with diastolic BP. Oscillometric data can also be interpreted for the purpose of obtaining information on arterial size as function of transmural pressure. Fig. 1 shows simulated examples of such arterial size - transmural pressure relationships, illustrating how arterial properties can vary from person to person and with time. The orange curve is illustrating an example of an artery under hemodynamic stability, while blue and green represent the possible state of the artery when hemodynamic compensation mechanisms are activated (vasodilation, respectively vasoconstriction). As shown by Bank et al. [17], administration of vasoactive drugs influences the relationship. Several studies have investigated methods of measuring the arterial cross-sectional area - transmural pressure relationship by means of oscillometric measurements [19], [20], [21]. Arterial properties such as compliance under a range of transmural pressures, as well as collapse characteristics can be obtained from this measurement.

2. Pulse wave velocity (PWV) measurements of arterial compliance are based on principles of pressure pulse wave propagation in compliant vs. rigid tubes. Pressure pulses generated by the heart travel faster along stiffer arteries than compliant arteries. Formulas such as Bramwell-Hill or models based on transmission line principles [22] are used to relate PWV to arterial compliance asPWV(Ptm)=V(Ptm)ρ*C(Ptm),where Ptm is transmural pressure across the arterial wall, ρ is the blood density, V(Ptm) is the arterial volume as a function of Ptm, and C(Ptm) is the arterial compliance as a function of Ptm.

PWV can be measured directly or it can be inferred from pulse travel time (i.e., the time of travel of the pressure pulse over a distance in the arterial tree). Applanation tonometry [34] or imaging technologies such as Doppler ultrasound and MRI have been used [23, 24]. A more accessible way to measure PWV which does not involve imaging, is based on measuring pulse arrival time PAT – the time delay between the R-peak of the electrocardiogram ECG signal and a fiducial point in the photoplethysmogram (PPG) signal at the finger site or other peripheral location.

Usual patient monitors include ECG, PPG and blood pressure cuffs. Therefore, it would be advantageous to infer brachial arterial compliance by assessment of arterial volume changes with respect to pressure changes through oscillometric measurements or by inference via pulse wave velocity derived from ECG/PPG based on data collected by the already available signals.

A source of uncertainty in the extraction of arterial properties from ECG/PPG-derived PWV is that vessel properties can differ significantly from central to peripheral locations. Therefore, assuming uniformity of the arterial properties is unrealistic. A complex transmission line model, together with the possibility to measure pulse arrival time at several locations between the heart at the finger PPG would be necessary to extract location-specific arterial properties. Drawbacks of measuring PWV based on ECG/PPG-derived PAT include the difficulty of approximating heart pre-ejection period (PEP), as well as measuring the exact distance that the pulse travels between the two sites.

A strategy to overcome these uncertainties can be derived from work concerning calibration techniques for BP surrogates [25]. Recording PAT changes during the inflation of a cuff placed on the upper arm enables the estimation of arterial properties specifically at the cuff location. This is due to the localized change in transmural pressure over the arterial wall at brachial site. As cuff pressure increases, the arterial transmural pressure over the length of the cuff is altered, thus changing the time it takes for the pulse to propagate down the artery. The velocity of the pulse is indicative of the artery characteristics for a range of transmural pressures.

Therefore, the measurement approach uses the cuff inflation to provide: 1) oscillometric data and 2) PAT changes induced by local changes in transmural pressure. A setup to simultaneously record oscillometric data and PAT based on this concept is shown in Fig. 2. A cuff is placed on the upper arm to obtain brachial artery volume oscillations from pressure oscillations in the cuff dependent on the transmural pressure. Simultaneously, PAT is recorded by means of computing the delay between the R-peak of the ECG signal waveform and a fiducial point of the PPG waveform (which is obtained at the finger site of the same arm as where the cuff is placed).

As a follow up of this experimental work [25], in this paper we present an algorithmic framework for joint processing of volume/pressure data and PWV changes to robustly infer changes in arterial compliance with the aim to monitor this hemodynamic parameter. The presented measurement method is designed to detect changes in arterial compliance from the brachial location. Arterial changes of systemic nature are detected at this site; this information complements the BP value extracted at the same brachial site through the cuff-based measurement. Our hypothesis is that this local information is also relevant for systemic changes in the vascular system.

Firstly, a forward model is presented which expresses the oscillometric and PAT data based on the mechanical properties of the artery. This is followed by introduction of a simulation framework to infer arterial compliance using a Bayesian fusion technique. The obtained results are discussed with emphasis on their clinical relevance.

Section snippets

Methodology

Information obtained from PAT and oscillometry measurements can be jointly processed by considering dynamics of arterial collapse and how such arterial mechanics are expressed in the two signals. More precisely, collapse mechanics are represented through a mathematical model describing arterial cross-sectional area changes as transmural pressure is varied [21]. The model parameters are expressed either in cuff pressure oscillations, in pulse wave velocity, or in both. The model reads as:A(Ptm)=d

Parameter inference from oscillometric data

An illustration of the oscillometric data simulated with parameter set a = 0.03, b = 0.1, c = 0.08 is shown is Fig. 9.

The simulated oscillometric data is fed into the MCMC sampling algorithm, which after a set number of iterations outputs posterior distributions for the three model parameters a, c and d (this process is illustrated in Fig. 7).

The posterior distribution can be summarized by the 95% highest density interval (HDI), which contains the 95% most probable parameter values, together

Discussion

This study presents an inference approach based on Bayesian principles for fusion of two measurements (oscillometry and PAT) with the goal of obtaining a robust estimate of arterial compliance in a realistic hemodynamic monitoring scenario with access to PAT and arterial volume envelope obtained from oscillometry. The inference modality is first demonstrated on the individual measurements via simulations (Fig. 10a, b, d). No artefacts are simulated in the oscillometric signal and expected

Conclusion

The presented simulation framework illustrates the feasibility of a data processing approach for obtaining information on arterial compliance by signal fusion of two data sources (PAT and oscillometry). The results show the basic feasibility of the concept, this being a first step towards improved patient monitoring via measurement of hemodynamic parameters related to arterial compliance. The framework allows for illustration of clinical scenarios and transparent interpretation on how the two

Declaration of Competing Interest

Jens Muehlsteff is an employee of Philips.

References (35)

  • B. Jiang et al.

    Measurement of Pulse Wave Velocity Using Pulse Wave Doppler Ultrasound: comparison with Arterial Tonometry

    Ultrasound Med Biol

    (2008)
  • S. Laurent et al.

    Expert consensus document on arterial stiffness: methodological issues and clinical applications

    Eur. Heart J.

    (2006)
  • R. Payne et al.

    Arterial Stiffness and Hypertension

    Hypertension

    (2010)
  • R. Townsend

    Arterial Stiffness: recommendations and Standardization

    Pulse

    (2016)
  • J.E. Hall et al.

    Guyton and Hall textbook of Medical Physiology

    (2011)
  • Nihon Kohden Europe. (2019). PWTT. [online] Available...
  • E.M. Goldberg et al.

    New Approaches to Evaluating and Monitoring Blood Pressure

    Curr Hypertens Rep

    (2016)
  • PULSE CONTOUR CARDIAC OUTPUT (PICCO) LEARNING PACKAGE. (2016). [online] Available...
  • J.-L. Vincent et al.

    Clinical review: Update on hemodynamic monitoring - a consensus of 16

    Crit Care

    (2011)
  • A.K. Khanna

    Defending a mean arterial pressure in the intensive care unit: are we there yet?

    Ann Intensive Care

    (2018)
  • E. Futier et al.

    Effect of Individualized vs Standard Blood Pressure Management Strategies on Postoperative Organ Dysfunction Among High-Risk Patients Undergoing Major Surgery

    JAMA

    (2017)
  • M. Leone et al.

    Optimizing mean arterial pressure in septic shock: a critical reappraisal of the literature

    Critical Care

    (2015)
  • P. Asfar et al.

    High versus low blood-pressure target in patients with septic shock

    N Engl J Med

    (2014)
  • M. Dünser et al.

    Re-thinking resuscitation: leaving blood pressure cosmetics behind and moving forward to permissive hypotension and a tissue perfusion-based approach

    Critical Care

    (2013)
  • G. Hernández et al.

    Is the macrocirculation really dissociated from the microcirculation in septic shock?

    Intensive Care Med

    (2016)
  • M. Cecconi et al.

    Consensus on circulatory shock and hemodynamic monitoring

    Task force of the European Society of Intensive Care Medicine. Intensive Care Medicine

    (2014)
  • G.E. McVeigh et al.

    Arterial compliance

    Cardiovascular Medicine

    (2007)
  • Cited by (0)

    View full text