Marcus Granegger

Development of a Pump Flow Estimator for Rotary Blood Pumps to Enhance Monitoring of Ventricular Function

Estimation of instantaneous flow in rotary blood pumps (RBPs) is important for monitoring the interaction between heart and pump and eventually the ventricular function. Our group has reported an algorithm to derive ventricular contractility based on the maximum time derivative (dQ/dt(max) as a substitute for ventricular dP/dt(max) ) and pulsatility of measured flow signals. However, in RBPs used clinically, flow is estimated with a bandwidth too low to determine dQ/dt(max) in the case of improving heart function. The aim of this study was to develop a flow estimator for a centrifugal pump with bandwidth sufficient to provide noninvasive cardiac diagnostics. The new estimator is based on both static and dynamic properties of the brushless DC motor. An in vitro setup was employed to identify the performance of pump and motor up to 20 Hz. The algorithm was validated using physiological ventricular and arterial pressure waveforms in a mock loop which simulated different contractilities (dP/dt(max) 600 to 2300 mm Hg/s), pump speeds (2 to 4 krpm), and fluid viscosities (2 to 4 mPa·s). The mathematically estimated pump flow data were then compared to the datasets measured in the mock loop for different variable combinations (flow ranging from 2.5 to 7 L/min, pulsatility from 3.5 to 6 L/min, dQ/dt(max) from 15 to 60 L/min/s). Transfer function analysis showed that the developed algorithm could estimate the flow waveform with a bandwidth up to 15 Hz (±2 dB). The mean difference between the estimated and measured average flows was +0.06 ± 0.31 L/min and for the flow pulsatilities -0.27 ± 0.2 L/min. Detection of dQ/dt(max) was possible up to a dP/dt(max) level of 2300 mm Hg/s. In conclusion, a flow estimator with sufficient frequency bandwidth and accuracy to allow determination of changes in ventricular contractility even in the case of improving heart function was developed.

Strong corruption of electrocardiograms caused by cardiopulmonary resuscitation reduces efficiency of two-channel methods for removing motion artefacts in non-shockable rhythms

AIM Cardiopulmonary resuscitation (CPR) artefact removal methods provide satisfactory results when the rhythm is shockable but fail on non-shockable rhythms. We investigated the influence of the corruption level on the performance of four different two-channel methods for CPR artefact removal. MATERIALS AND METHODS 395 artefact-free ECGs and 13 pure CPR artefacts with corresponding blood pressure readings as a reference channel were selected. Using a simplified additive data model we generated CPR-corrupted signals at different signal-to-noise ratio (SNR) levels from -10 to +10 dB. The algorithms were optimized on learning data with respect to SNR improvement and then applied to testing data. Sensitivity and specificity were derived from the shock/no-shock advice of an automated external defibrillator before CPR corruption and after artefact removal. RESULTS Sensitivity for the filtered data (>95%) was significantly superior to that for the unfiltered data (76%), p<0.001. However, specificity was similar for the filtered and unfiltered data (<90% vs 89.3%). For large artefacts (-10 dB) specificity decreased below 70%. No important difference in the performance of the four algorithms was found. CONCLUSION Using a simplified data model we showed that, when the ECG rhythm is non-shockable, two-channel methods could not reduce CPR artefacts without affecting the rhythm analysis for shock recommendation. The reason could be poor reconstruction when the artefacts are large. However, poor reconstruction was not a hindrance to re-identifying shockable rhythms. Future investigations should both include the refinement of filter methods and also focus on reducing motion artefacts already at the recording stage.

Assessment of Aortic Valve Opening During Rotary Blood Pump Support Using Pump Signals

During left ventricular support by rotary blood pumps (RBPs), the biomechanics of the aortic valve (AV) are altered, potentially leading to adverse events like commissural fusion, valve insufficiency, or thrombus formation. To avoid these events, assessment of AV opening and consequent adaptation of pump speed seem important. Additionally, this information provides insight into the heart-pump interaction. The aim of this study was to develop a method to assess AV opening from the pump flow signal. Data from a numerical model of the cardiovascular system and animal experiments with an RBP were employed to detect the AV opening from the flow waveform under different hemodynamic conditions. Three features calculated from the pump flow waveform were used to classify the state of the AV: skewness, kurtosis, and crest factor. Three different classification algorithms were applied to determine the state of the AV based on these features. In the model data, the best classifier resulted in a percentage of correctly identified beats with a closed AV (specificity) of 99.9%. The percentage of correctly identified beats with an open AV (sensitivity) was 99.5%. In the animal experiments, specificity was 86.8% and sensitivity reached 96.5%. In conclusion, a method to detect AV opening independently from preload, afterload, heart rate, contractility, and degree of support was developed. This algorithm makes the evaluation of the state of the AV possible from pump data only, allowing pump speed adjustment for a frequent opening of the AV and providing information about the interaction of the native heart with the RBP.

Evaluation of Left Ventricular Relaxation in Rotary Blood Pump Recipients Using the Pump Flow Waveform: A Simulation Study

In heart failure, diastolic dysfunction is responsible for about 50% of the cases, with higher prevalence in women and elderly persons and contributing similarly to mortality as systolic dysfunction. Whereas the cardiac systolic diagnostics in ventricular assist device patients from pump parameters have been investigated by several groups, the diastolic behavior has been barely discussed. This study focuses on the determination of ventricular relaxation during early diastole in rotary blood pump (RBP) recipients. In conventional cardiology, relaxation is usually evaluated by the minimum rate and the time constant of left ventricular pressure decrease, dP/dt(min) and τ(P) . Two new analogous indices derived from the pump flow waveform were investigated in this study: the minimum rate and the time constant of pump flow decrease, dQ/dt(min) and τ(Q) . The correspondence between the indices was investigated in a numerical simulation of the assisted circulation for different ventricular relaxation states (τ(P) ranging from 24 to 68 ms) and two RBP models characterized by linear and nonlinear pressure-flow characteristics. dQ/dt(min) and τ(Q) always correlated with the dP/dt(min) and τ(P) , respectively (r>0.97). These relationships were influenced by the nonlinear pump characteristics during partial support and by the pump speed during full support. To minimize these influences, simulation results suggest the evaluation of dQ/dt(min) and τ(Q) at a pump speed that corresponds to the borderline between partial and full support. In conclusion, at least in simulation, relaxation can be derived from pump data. This noninvasively accessible information could contribute to a continuous estimation of the remaining cardiac function and its eventual recovery.

Continuous Monitoring of Cardiac Rhythms in Left Ventricular Assist Device Patients

Monitoring of cardiac rhythms is of major importance in the treatment of heart failure patients with left ventricular assist devices (LVADs) implanted. A continuous surveillance of these rhythms could improve out-of-hospital care in these patients. The aim of this study was to investigate cardiac rhythms using available pump data only. Datasets (n = 141) obtained in the normal ward, in the intensive care unit, and during bicycle ergometry were analyzed in 11 recipients of a continuous flow LVAD (59.1 ± 9.7 years; male 82%). Tachograms and arrhythmic patterns derived from the pump flow waveform, and a simultaneously recorded ECG were compared, as well as heart rate variability parameters such as: the average heart beat duration (RR interval), the standard deviation of the beat duration (SDNN), the root-mean-square of the difference of successive beat durations (RMSSD), and the number of pairs of adjacent beat duration differing by >50 ms divided by the number of all beats (pNN50). A very good agreement of cardiac rhythm parameters from the pump flow compared with ECG was found. Tachycardia, atrial fibrillation, and extrasystoles could be accurately identified from the tachograms derived from the pump flow. Also, Bland-Altman analysis comparing pump flow with ECG indicated a very small difference in average RR interval of 0.3 ± 1.0 ms, in SSDN of 0.5 ± 2.7 ms, in RMSSD of 1.0 ± 5.6 ms, and in pNN50 of 0.3 ± 1.0%. Continuous monitoring of cardiac rhythms from available pump data is possible. It has the potential to reduce the out-of-hospital diagnostic burden and to permit a more efficient adjustment of the level of mechanical support.

Use of continuous flow ventricular assist devices in patients with heart failure and a normal ejection fraction: A computer-simulation study

OBJECTIVES Continuous flow left ventricular assist devices are used in end-stage systolic heart failure. However, about one half of the patients with heart failure exhibit diastolic dysfunction with a normal ejection fraction. In the present study, the possible hemodynamic consequences of continuous flow left ventricular assist devices use for these patients were investigated. METHODS A previously developed cardiovascular model was modified to reproduce the peculiar hemodynamics of heart failure with a normal ejection fraction. The model was based on and validated with patient data derived from the published data. A continuous flow left ventricular assist device model was included and the hemodynamic effects of pump support evaluated at rest and during exercise. RESULTS The model accurately reproduced the published data both at rest and during exercise, leading to simulated hemodynamic values within the standard deviations of patient variability. At rest, pump support decreased the end-diastolic left ventricular pressure (6 vs 15 mm Hg) and volume (88 vs 135 mL). During exercise, maximal pump support substantially unloaded the left ventricle (end-diastolic pressure, 14 vs 35 mm Hg; volume, 133 vs 158 mL) and the pulmonary venous circulation (left atrial pressure, 12 vs 24 mm Hg) and resulted in a slight increase in cardiac output (11.7 vs 9.9 L/min). CONCLUSIONS The simulation results suggested that continuous flow left ventricular assist devices improve the hemodynamics in patients with heart failure and a normal ejection fraction. For an optimal use of continuous flow left ventricular assist devices, low speeds should be maintained at rest, to avoid suction. However, during physical activity, higher speeds are needed to prevent an abnormal increase in the ventricular filling pressures typical of patients with heart failure and a normal ejection fraction.

Continuous Monitoring of Aortic Valve Opening in Rotary Blood Pump Patients

Goal: Rotary blood pumps (RBPs) typically support the left ventricle by pumping blood from the ventricle to the aorta, partially bypassing the aortic valve (AV). Monitoring the AV opening during RBP support would provide important information about cardiac-pump interaction. However, currently this information is not continuously available. In this study, an algorithm to determine AV opening using available pump signals was evaluated in humans. Methods: Pump speed changes were performed in 15 RBP patients to elicit opening of the AV. Simultaneously to pump data recordings, the AV was continuously monitored using echocardiography. The algorithm, which classifies the AV state utilizing three features (skewness, kurtosis, and crest factor) calculated from the pump flow waveform, was compared to echocardiography by using cross-validation analysis. Additionally, numerical simulation was used to evaluate effects of different pump characteristics and cannula length, as well as mitral valve insufficiency on the AV opening detection method. Results: More than 7000 heart beats were analyzed. The correct classification rate using the developed algorithm was 91.1% (sensitivity 91.0%, specificity 91.2%). Numerical simulations showed that the flow waveform shape used for AV opening detection is preserved under the different conditions studied. Conclusion: This study demonstrates that the AV opening can be reliably detected in RBP patients using available pump data. Significance: Once implemented in RBP controllers, this method will provide a novel tool to improve the management of RBP patients, particularly for adjustments of the pump speed and flow and for the evaluation of the assisted cardiac function.

Investigation of hemodynamics in the assisted isolated porcine heart

Background Currently, the interaction between rotary blood pumps (RBP) and the heart is investigated in silico, in vitro, and in animal models. Isolated and defined changes in hemodynamic parameters are unattainable in animal models, while the heart-pump interaction in its whole complexity cannot be modeled in vitro or in silico. Aim The aim of this work was to develop an isolated heart setup to provide a realistic heart-pump interface with the possibility of easily adjusting hemodynamic parameters. Methods A mock circuit mimicking the systemic circulation was developed. Eight porcine hearts were harvested using a protocol similar to heart transplantation. Then, the hearts were resuscitated using Langendorff perfusion with rewarmed, oxygenated blood. An RBP was implanted and the setup was switched to the “working mode” with the left heart and the RBP working as under physiologic conditions. Both the unassisted and assisted hemodynamics were monitored. Results In the unassisted condition, cardiac output was up to 9.5 L/min and dP/dtmax ranged from 521 to 3621 mmHg/s at a preload of 15 mmHg and afterload of 70 mmHg. With the RBP turned on, hemodynamics similar to heart-failure patients were observed in each heart. Mean pump flow and flow pulsatility ranged from 0 to 11 L/min. We were able to reproduce conditions with an open and closed aortic valve as well as suction events. Conclusions An isolated heart setup including an RBP was developed, which combines the advantages of in silico/vitro methods and animal experiments. This tool thus provides further insight into the interaction between the heart and an RBP.

Human ECGs corrupted with real CPR artefacts in an animal model: Generating a database to evaluate and refine algorithms for eliminating CPR artefacts

AIM For the analysis of ECG rhythms during ongoing CPR, single- or two-channel methods have been proposed to eliminate artefacts from the CPR-corrupted ECG. To refine, test and evaluate these algorithms with a realistic data set, we introduce an animal model with which we created an extended database of human ECGs with real CPR artefacts. MATERIAL AND METHODS In a pig model real CPR-related artefacts were added to annotated human emergency ECGs. Via a special catheter placed in the oesophagus, ECG sequences (duration>10s) were fed in close to the dead pigs heart. The resulting surface potential was recorded on the thorax without and during ongoing chest compressions, which were monitored using a miniature force sensor. RESULTS The animals served as a vehicle for human ECGs, making it possible to create a database in which 918 real human ECG sequences (437 shockable and 481 non-shockable) were corrupted with CPR-induced artefacts. The achieved signal-to-noise ratios (SNR) ranged from -17 to +15 dB, sensitivity was 93.5% and specificity was 50.51%. The fed-in ECG and the uncorrupted surface ECG correlated almost perfectly (r=0.926+/-0.081; n=918), indicating negligible signal distortion due to the dead pig itself. CONCLUSION As the generated database includes both the original and the corrupted ECG covering a wide range of SNRs as well as the compression force signal, it provides an extended data set to evaluate the reconstruction performance of CPR artefact-removal algorithms.

Blood damage in ventricular assist devices

Advancements in both the design of mechanical circulatory support (MCS) and medical care for recipients have led to steady improvements in patient survival over the last decades. Indeed, current generations of ventricular assist devices (VADs), for example, provide important therapeutic alternatives for individuals with heart failure that may be ineligible for heart transplant. Nevertheless, it is clear that freedom from adverse events (including premature death) is experienced by an unacceptably low proportion of MCS recipients, as evidenced by recent INTERMACS reports.1 The primary complications observed post surgery remain neurological disorders (e.g. stroke), multi-system organ failure and infection. While infection remains an ongoing area of interest with important advancement, it is currently being explored whether a common aetiology may explain the high rate of organ failure (including the brain).