In vitro evaluation of multiobjective hemodynamic control of a heart-assist pump.

Ventricular assist devices now clinically used for treatment of end-stage heart failure require responsive and reliable hemodynamic control to accommodate the continually changing demands of the body. This is an essential ingredient to maintaining a high quality of life. To satisfy this need, a control algorithm involving a trade-off between optimal perfusion and avoidance of ventricular collapse has been developed. An optimal control strategy has been implemented in vitro that combines two competing indices: representing venous return and prevalence of suction. The former is derived from the first derivative of diastolic flow with speed, and the latter derived from the harmonic spectra of the flow signal. The responsiveness of the controller to change in preload and afterload were evaluated in a mock circulatory simulator using a HeartQuest centrifugal blood pump (CF4b, MedQuest Products, Salt Lake City, UT). To avoid the need for flow sensors, a state estimator was used, based on the back-EMF of the actuator. The multiobjective algorithm has demonstrated more robust performance as compared with controllers relying on individual indices.

A suction detection system for rotary blood pumps based on the Lagrangian support vector machine algorithm.

The Left Ventricular Assist Device (LVAD) is a rotary mechanical pump that is implanted in patients with congestive heart failure to help the left ventricle in pumping blood in the circulatory system. However, using such a device may result in a very dangerous event, called ventricular suction that can cause ventricular collapse due to overpumping of blood from the left ventricle when the rotational speed of the pump is high. Therefore, a reliable technique for detecting ventricular suction is crucial. This paper presents a new suction detection system that can precisely classify pump flow patterns, based on a Lagrangian Support Vector Machine (LSVM) model that combines six suction indices extracted from the pump flow signal to make a decision about whether the pump is in suction, approaching suction, or not in suction. The proposed method has been tested using in vivo experimental data based on two different pumps. The simulation results show that the system can produce superior performance in terms of classification accuracy, stability, learning speed, and good robustness compared to three other existing suction detection methods and the original SVM-based algorithm. The ability of the proposed algorithm to detect suction provides a reliable platform for the development of a feedback control system to control the speed of the pump while at the same time ensuring that suction is avoided.

An engineering analysis of the aortic valve dynamics in patients with rotary Left Ventricular Assist Devices.

The use of a rotary Left Ventricular Assist Device (LVAD) as a bridge-to-recovery treatment is gaining considerable attention in the LVAD research community. Using a mathematical model of the cardiovascular-LVAD system, this paper intends to define the critical control parameters in terms of power and rotational speed of the LVAD to ensure normal dynamics of the aortic valve for different levels of patient's activity and severity of heart failure. The effects of permanent closure of the aortic valve on the hemodynamics of the patient and the pump flow characteristics, if the critical control values are exceeded, are also examined. Additionally, LVAD power and speed control parameters that yield a given percentage of the cardiac cycle during which the aortic valve remains open are examined indicating that the severity of the heart failure is a very important factor in deciding the appropriateness of the LVAD as a bridge-to recovery treatment.

Detection of ventricular suction in an implantable rotary blood pump using support vector machines.

A new suction detection algorithm for rotary Left Ventricular Assist Devices (LVAD) is presented. The algorithm is based on a Lagrangian Support Vector Machine (LSVM) model. Six suction indices are derived from the LVAD pump flow signal and form the inputs to the LSVM classifier. The LSVM classifier is trained and tested to classify pump flow patterns into three states: No Suction, Approaching Suction, and Suction. The proposed algorithm has been tested using existing in vivo data. When compared to three existing methods, the proposed algorithm produced superior performance in terms of classification accuracy, stability, and learning speed. The ability of the algorithm to detect suction provides a reliable platform in the development of a pump speed controller that has the capability of avoiding suction.

A method for subsample fetal heart rate estimation under noisy conditions.

In this paper, we consider a new approach for estimating the fundamental period in fetal ECG waveforms. The fundamental period contains information that is indicative of the physiological condition of the fetus such as hypoxia and acidemia. Our method is based on the minimization of a cost function that measures the differences between the discrete Fourier transform (DFT) of the fetal ECG waveform and the DFTs of its circularly shifted forms. By using the linear phase shift property of the DFT, we show that the minimization of this cost function is equivalent to finding the cosine waveform that matches best to the ECG power spectrum. The optimal cosine waveform is then used to estimate the fundamental period. We expand this method and discuss estimation of the fundamental period with subsample precision. Subsample estimates may be useful especially when a low sampling rate is used for a long period of monitoring. Comparison of performance of this method with Cepstrum and average mIn this paper, we consider a new approach for estimating the fundamental period in fetal ECG waveforms. The fundamental period contains information that is indicative of the physiological condition of the fetus such as hypoxia and acidemia. Our method is based on the minimization of a cost function that measures the differences between the discrete Fourier transform (DFT) of the fetal ECG waveform and the DFTs of its circularly shifted forms. By using the linear phase shift property of the DFT, we show that the minimization of this cost function is equivalent to finding the cosine waveform that matches best to the ECG power spectrum. The optimal cosine waveform is then used to estimate the fundamental period. We expand this method and discuss estimation of the fundamental period with subsample precision. Subsample estimates may be useful especially when a low sampling rate is used for a long period of monitoring. Comparison of performance of this method with Cepstrum and average magnitude difference function methods shows that our approach achieves very accurate period estimation results for both simulated and real fetal EGC waveforms that are taken at different stages of the gestation under noisy conditions.agnitude difference function methods shows that our approach achieves very accurate period estimation results for both simulated and real fetal EGC waveforms that are taken at different stages of the gestation under noisy conditions.

Performance prediction of a percutaneous ventricular assist system using nonlinear circuit analysis techniques.

A percutaneous ventricular assist device (pVAD) is an extracorporeal cardiac assist system that supports the failing ventricle in advanced stage heart failure by bypassing blood from the venous to the arterial circulation through a blood pump. The system can be implanted in a Cath lab using standard interventional techniques, and typically consists of a venous or atrial drainage cannula, the VAD (or blood pump), and an arterial perfusion cannula. Because the device allows clinicians the freedom of choosing the configuration and size of the cannulae based on the patient's body size and the size of the artery, it is extremely difficult but important to be able to predict the amount of blood flow that the device can provide before it is implanted to support the patient. In this paper, we develop a novel method that can be used to accurately predict the mean flow rate that the device can provide to the patient based on the size and configuration of the arterial cannula, the pump speed, and the patient's left atrial and mean arterial pressures. To do this, we first develop a nonlinear electric circuit model for the pVAD. This model includes a speed dependent voltage source and flow dependent resistors to simulate the pressure-flow relationship in the various cannulae in the device. We show that the flow rate through the device can be determined by solving a quadratic equation whose coefficients are scaled depending on the size and configuration of the arterial cannula. The model and prediction method were tested experimentally on a test loop supported by the TandemHeart pVAD (Cardiacassist, Inc., Pittsburgh, PA). A comparison of the predicted flow rates obtained from our method with experimental data shows that our method can predict the flow rates accurately with error indices less than 6% for all test conditions over the entire range of intended use of the device. Computer simulations of the pVAD model coupled to a cardiovascular model showed that the accuracy- of the method in estimating the mean flow rate is consistent over the normal range of operation of the device regardless of the pulsatility introduced by the cardiovascular system. This method can be used as an additional too to assist cardiologists in choosing a proper arterial cannulae configurations and sizes for pVAD patients. It can also be used as a tool to train clinical personnel to operate the device under different physiological conditions.

A discriminant-analysis-based suction detection system for rotary blood pumps.

A new suction detection system for rotary blood pumps used in left ventricular assist devices is presented. The system can correctly classify pump flow patterns, based on a discriminant analysis (DA) model that combines several indices derived from the pump flow signal to make a decision about the pump status. The indices considered in this approach are frequency-, time-, and time-frequency-domain indices. The frequency-domain indices detect changes in the harmonic and subharmonic energy content of the pump flow signal when a suction event is occurring. The time-domain indices detect changes in pump flow pulsatility based on a beat-to-beat analysis of the pump flow and first derivative of pump flow. The time-frequency index can track variations in the standard deviation of the instantaneous frequency of the pump flow signal. These indices are combined in a DA decision system to generate a suction alarm. The proposed system has been tested in simulations and in-vivo experimental tests and produced satisfactory results.