Abstract

Non-invasive detection of dangerous pumping states is important for the control of rotary blood pumps. The VentrAssistTM LVAD, is an implantable centrifugal rotary blood pump (iRBP), the impeller supported by a hydrodynamic bearing. In-vitro (pulsatile mock-loop) and ex-vivo (acute sheep) experiments (N=3) were conducted. By perturbing system parameters of the pulsatile mock loop and by physiological intervention through pharmacological or mechanical methods in the ovine experiments, and by varying iRBP impeller speed, it was possible to transition through five physiologically significant pumping states. These states were defined as regurgitant pump flow, ventricular ejection, continuous aortic valve closure, partial sucking of the ventricle, and total ventricular collapse, states confirmed in the experiments through observation and through implanting of additional flow and pressure sensors. Using instantaneous impeller speed and pump input power, we investigated the detection of these states without the use of additional sensors. Indices based on pump flow, flow amplitude and flow waveform symmetry were analyzed using classification and regression tree methods. The classification tree derived from the in-vitro experiments was applied to the ex-vivo data with perfect accuracy (0% false positives and negatives in predicting the pumping states). These data indicate that the state detection methods may be a valuable mechanism in optimal LVAD control and operational diagnosis

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call