Guanqun Zhang

Pulse arrival time is not an adequate surrogate for pulse transit time as a marker of blood pressure

Pulse transit time (PTT) is a proven, simple to measure, marker of blood pressure (BP) that could potentially permit continuous, noninvasive, and cuff-less BP monitoring (after an initial calibration). However, pulse arrival time (PAT), which is equal to the sum of PTT and the pre-ejection period, is gaining popularity for BP tracking, because it is even simpler to measure. The aim of this study was to evaluate the hypothesis that PAT is an adequate surrogate for PTT as a marker of BP. PAT and PTT were estimated through the aorta using high-fidelity invasive arterial waveforms obtained from six dogs during wide BP changes induced by multiple interventions. These time delays and their reciprocals were evaluated in terms of their ability to predict diastolic, mean, and systolic BP (DBP, MBP, and SBP) per animal. The root mean squared error (RMSE) between the BP parameter predicted via the time delay and the measured BP parameter was specifically used as the evaluation metric. Taking the reciprocals of the time delays tended to reduce the RMSE values. The DBP, MBP, and SBP RMSE values for 1/PAT were 9.8 ± 5.2, 10.4 ± 5.6, and 11.9 ± 6.1 mmHg, whereas the corresponding values for 1/PTT were 5.3 ± 1.2, 4.8 ± 1.0, and 7.5 ± 2.2 mmHg (P < 0.05). Thus tracking BP via PAT was not only markedly worse than via PTT but also unable to meet the FDA BP error limits. In contrast to previous studies, our results quantitatively indicate that PAT is not an adequate surrogate for PTT in terms of detecting challenging BP changes.

Improved pulse transit time estimation by system identification analysis of proximal and distal arterial waveforms

We investigated the system identification approach for potentially improved estimation of pulse transit time (PTT), a popular arterial stiffness marker. In this approach, proximal and distal arterial waveforms are measured and respectively regarded as the input and output of a system. Next, the system impulse response is identified from all samples of the measured input and output. Finally, the time delay of the impulse response is detected as the PTT estimate. Unlike conventional foot-to-foot detection techniques, this approach is designed to provide an artifact robust estimate of the true PTT in the absence of wave reflection. The approach is also applicable to arbitrary types of arterial waveforms. We specifically applied a parametric system identification technique to noninvasive impedance cardiography (ICG) and peripheral arterial blood pressure waveforms from 15 humans subjected to lower-body negative pressure. We assessed the technique through the correlation coefficient (r) between its 1/PTT estimates and measured diastolic pressure (DP) per subject and the root mean squared error (RMSE) of the DP predicted from these estimates and measured DP. The technique achieved average r and RMSE values of 0.81 ± 0.16 and 4.3 ± 1.3 mmHg. For comparison, the corresponding values were 0.59 ± 0.37 (P < 0.05) and 5.9 ± 2.5 (P < 0.01) mmHg for the conventional technique applied to the same waveforms and 0.28 ± 0.40 (P < 0.001) and 7.2 ± 1.8 (P < 0.001) mmHg for the conventional technique with the ECG waveform substituted for the ICG waveform. These results demonstrate, perhaps for the first time, that the system identification approach can indeed improve PTT estimation.

Tube-load model parameter estimation for monitoring arterial hemodynamics.

A useful model of the arterial system is the uniform, lossless tube with parametric load. This tube-load model is able to account for wave propagation and reflection (unlike lumped-parameter models such as the Windkessel) while being defined by only a few parameters (unlike comprehensive distributed-parameter models). As a result, the parameters may be readily estimated by accurate fitting of the model to available arterial pressure and flow waveforms so as to permit improved monitoring of arterial hemodynamics. In this paper, we review tube-load model parameter estimation techniques that have appeared in the literature for monitoring wave reflection, large artery compliance, pulse transit time, and central aortic pressure. We begin by motivating the use of the tube-load model for parameter estimation. We then describe the tube-load model, its assumptions and validity, and approaches for estimating its parameters. We next summarize the various techniques and their experimental results while highlighting their advantages over conventional techniques. We conclude the review by suggesting future research directions and describing potential applications.

Improved Pulse Wave Velocity Estimation Using an Arterial Tube-Load Model

Pulse wave velocity (PWV) is the most important index of arterial stiffness. It is conventionally estimated by noninvasively measuring central and peripheral blood pressure (BP) and/or velocity (BV) waveforms and then detecting the foot-to-foot time delay between the waveforms wherein wave reflection is presumed absent. We developed techniques for improved estimation of PWV from the same waveforms. The techniques effectively estimate PWV from the entire waveforms, rather than just their feet, by mathematically eliminating the reflected wave via an arterial tube-load model. In this way, the techniques may be more robust to artifact while revealing the true PWV in absence of wave reflection. We applied the techniques to estimate aortic PWV from simultaneously and sequentially measured central and peripheral BP waveforms and simultaneously measured central BV and peripheral BP waveforms from 17 anesthetized animals during diverse interventions that perturbed BP widely. Since BP is the major acute determinant of aortic PWV, especially under anesthesia wherein vasomotor tone changes are minimal, we evaluated the techniques in terms of the ability of their PWV estimates to track the acute BP changes in each subject. Overall, the PWV estimates of the techniques tracked the BP changes better than those of the conventional technique (e.g., diastolic BP root-mean-squared errors of 3.4 versus 5.2 mmHg for the simultaneous BP waveforms and 7.0 versus 12.2 mmHg for the BV and BP waveforms (p <; 0.02)). With further testing, the arterial tube-load model-based PWV estimation techniques may afford more accurate arterial stiffness monitoring in hypertensive and other patients.

Robust, beat-to-beat estimation of the true pulse transit time from central and peripheral blood pressure or flow waveforms using an arterial tube-load model

We proposed a technique for estimating beat-to-beat pulse transit time (PTT) from central and peripheral blood pressure or flow waveforms based on an arterial tube-load model of wave reflection. The technique effectively estimates PTT from the entire waveforms after mathematically eliminating the reflected wave. So, unlike the conventional foot-to-foot detection technique, this technique should be robust to artifact while revealing the true PTT (i.e., the PTT in absence of wave reflection). We compared the two techniques, as applied to blood pressure and flow waveforms, in terms of the ability of their PTT estimates to correlate with blood pressure (a) during baseline (for which the naturally occurring beat-to-beat changes were small), (b) during low heart rate (wherein wave reflection was profound), and (c) in the presence of actual measurement artifact. In all three cases, the PTT estimates of the arterial tube-load model technique yielded markedly superior correlation to blood pressure.

Assessing the challenges of a pulse wave velocity based blood pressure measurement in surgical patients.

Development of a continuous noninvasive blood pressure (cNIBP) monitor that is unobtrusive to patients is an attractive alternative to the cuff based measurements performed on medical-surgical floors in the hospital. Pulse wave velocity (PWV) provides a means to continuously monitor blood pressure in these patients. However, a PWV based cNIBP monitor faces a number of challenges in order to accurately measure blood pressure. In our study, we investigated some of the challenges faced by a body-worn cNIBP monitor (i.e. ViSi Mobile) on data collected on patients undergoing surgery. Results indicated that 1) pulse arrival time (PAT) values from ViSi Mobile were well correlated with PAT values obtained from an invasive reference; 2) the reciprocal of the PAT measurements were linearly correlated with blood pressure but the calibration curve was altered by administration of certain vasoactive substances; and 3) there are deterministic correlations between systolic pressure, diastolic pressure and the corresponding mean arterial pressure over a wide range of blood pressure values.

Robust pulse wave velocity estimation by application of system identification to proximal and distal arterial waveforms

Pulse wave velocity (PWV) is a marker of arterial stiffness and may permit continuous, non-invasive, and cuff-less monitoring of blood pressure. Here, robust PWV estimation was sought by application of system identification to proximal and distal arterial waveforms. In this approach, the system that optimally couples the proximal waveform to the distal waveform is identified, and the time delay of this system is then used to calculate PWV. To demonstrate proof-of-concept, a standard identification technique was applied to non-invasive impedance cardiography and peripheral arterial blood pressure waveforms from six humans subjected to progressive reductions in central blood volume induced by lower body negative pressure. This technique estimated diastolic pressure with an overall root-mean-squared-error of 5.2 mmHg. For comparison, the conventional detection method for estimating PWV yielded a corresponding error of 8.3 mmHg.

Monitoring aortic stiffness in the presence of measurement artifact based on an arterial tube model

Pulse wave velocity (PWV) determined through the foot-to-foot time delay between carotid and femoral artery waveforms is an index of aortic stiffness with proven clinical value. However, handheld transducers, which are often used to non-invasively measure the waveforms, are prone to motion artifact that may limit the full potential of this index. Here, we conceived an artifact robust technique to estimate PWV based on an arterial tube model. We applied the technique to high fidelity canine arterial pressure waveforms before and after contamination with known amounts of noise. Our results showed that, as the signal-to-noise ratio decreased, the PWV estimates of the technique predicted diastolic and mean arterial pressure with increasingly greater accuracy than the PWV estimates of the conventional foot-to-foot detection technique.

Early detection of hemorrhage via central pulse pressure derived from a non-invasive peripheral arterial blood pressure waveform

There is a profound need for early and convenient detection of hemorrhage in both civilian and military medicine. Due to wave reflection timing, central pulse pressure (PP), but not peripheral PP, is a surrogate of stroke volume (SV) and therefore an early marker of blood loss. However, only peripheral PP is convenient to measure. We tested an adaptive transfer function technique for deriving the central arterial blood pressure (ABP) waveform from a non-invasive peripheral ABP waveform in healthy humans subjected to lower body negative pressure (LBNP), a safe model of early hemorrhage. Our results showed that the derived central PP provided an earlier and more sensitive marker of progressive LBNP and a far more accurate measure of SV than measured peripheral PP.

Pulse arrival time is not an adequate surrogate for pulse transit time in terms of tracking diastolic pressure

We compared pulse arrival time (PAT) and pulse transit time (PTT) in terms of their ability to track diastolic pressure (DP). We performed the comparison using high fidelity, invasive arterial waveforms recorded from six dogs during multiple interventions. On average, DP ranged from 40 to 106 mmHg and therefore varied widely. PAT and PTT were able to predict DP with average root-mean-squared-errors of 9.8±5.8 mmHg and 5.7±2.0 mmHg (p = 0.02). Thus, even though PAT is simpler to measure, we can only recommend using PTT for tracking DP.