Search

In this study, a tripod supported sealless centrifugal blood pump was designed and fabricated for implantable application using a specially designed DC brushless motor. The tripod structure consists of 3 ceramic balls mounted at the bottom surface of the impeller moving in a polyethylene groove incorporated at the bottom pump casing. The follower magnet inside the impeller is coupled to the driver magnet of the motor outside the bottom pump casing, thus allowing the impeller to slide-rotate in the polyethylene groove as the motor turns. The pump driver has a weight of 230 g and a diameter of 60 mm. The acrylic pump housing has a weight of 220 g with the priming volume of 25 ml. At the pump rpm of 1,000 to 2,200, the generated head pressure ranged from 30 to 150 mm Hg with the maximum system efficiency being 12%. When the prototype pump was used in the pulsatile mock loop to assist the ventricle from its apex to the aorta, a strong correlation was obtained between the motor current and bypass flow waveforms. The waveform deformation index (WDI), defined as the ratio of the fundamental to the higher order harmonics of the motor current power spectral density, was computed to possibly detect the suction occurring inside the ventricle due to the prototype centrifugal pump. When the WDI was kept under the value of 0.20 by adjusting the motor rpm, it was successful in suppressing the suction due to the centrifugal pump in the ventricle. The prototype sealless, centrifugal pump together with the control method based on the motor current waveform analysis may offer an intermediate support of the failing left or right ventricle bridging to heart transplantation.

To measure blood hematocrit inside the magnetically suspended centrifugal pump, we have performed both forward and backward light scattering measurements using a specially designed optical cell. In the forward scattering measurement, an optical fiber was used to guide the near infrared light at 780 nm into a 250 microns gap region, and the light that forward scattered toward a detector fiber was measured using a phototransistor. The light intensity decreased exponentially with an increase in the hematocrit to around 20%. The forward scattering method suffered from sensitivity at the hematocrit levels around 25-45% due to the diffusion effect. By making the optical path length larger than several millimeters, the sensitivity of the forward scattering method in terms of hematocrit change can be improved. In the back scattering method, however, better sensitivity in terms of hematocrit change from 0-50% was obtained. By making the optical fiber separation distance less than 1 mm, the system will measure the first order back scattering from the shallow layer while, by making the fiber separation distance larger than several millimeters, the system will primarily measure the diffuse reflectance from the deeper layer. Both approaches will yield sensitive optical intensity change in terms of the physiological hematocrit range.

In this study, a detection algorithm for suction and regurgitation of the centrifugal pump during left heart bypass without relying on external flow or pressure sensors was developed and evaluated in acute studies using adult goats. The detection scheme relies on power spectral density (PSD) analysis of the motor current waveform through which the waveform deformation index (WDI) is obtained. This index is defined as the ratio of the fundamental component of the PSD to the higher PSD components, and its value increases with the deformation of the basic waveform. By assuming that the undistorted motor current waveform can be represented by a pure sine waveform, we theoretically synthesized various waveforms which have different second harmonic components. We were able to synthesize the waveform whose shape was close to the distorted motor current waveform under varying suction levels obtained in a mock loop study. From this study, we came to the conclusion that the WDI value of 0.2 can serve as a threshold level in deciding the suction and regurgitation speeds (rpm) during left heart bypass. In the study using adult goats, we were successful in minimizing both regurgitation and suction when the centrifugal pump speed was adjusted based on the WDI algorithm. The resultant bypass flow ranged from 1.5 to 2.0 L/min which was around 60% of the total flow. Further study is underway to evaluate the applicability of the WDI method in optimizing bypass pump flow.

An ultracompact, completely implantable permanent use electromechanical ventricular assist device (VAD) and total artificial heart (TAH) intended for 50-60 kg size patients have been developed. The TAH and VAD share a miniature electromechanical actuator that comprises a DC brushless motor and a planetary roller screw. The rotational force of the motor is converted into the rectilinear force of the roller screw to actuate the blood pump. The TAH is a one piece design with left and right pusher plate type blood pumps sandwiching an electromechanical actuator. The VAD is one half of the TAH with the same actuator but a different pump housing and a backplate. The blood contacting surfaces, including those of the flexing diaphragm and pump housing, of both the VAD and TAH were made of biocompatible polyurethane. The diameter, thickness, volume, and weight of the VAD are 90 mm, 56 mm, 285 cc, and 380 g, respectively, while those of the TAH are 90 mm, 73 mm, 400 cc, and 440 g, respectively. The design stroke volume of both the VAD and TAH is 60 cc with the stroke length being 12 mm. The stroke length and motor speed are controlled solely based on the commutation signals of the motor. An in vitro study revealed that a maximum pump flow of 7.5 L/min can be obtained with a pump rate of 140 bpm against a mean afterload of 100 mm Hg. The power requirement ranged from 4 to 6 W to deliver a 4-5 L/min flow against a 100 mm Hg afterload with the electrical-to-hydraulic efficiency being 19-20%. Our VAD and TAH are the smallest of the currently available devices and suitable for bridge to transplant application as well as for permanent circulatory support of 50-60 kg size patients.

In this study, the power output and contraction length of trained and untrained canine latissimus dorsi (LD) muscles were measured using a roller screw linear muscle actuator (RSLMA). The RSLMA consisted of a roller screw-nut assembly and translation unit to convert the linear pull force of the muscle into an axial displacement of the roller screw. When a cable wound around a spool attached at each end of the roller screw nut was pulled in either direction, the nut was rotated which in turn advanced the roller screw in the axial direction. Under anesthesia, a Telectronics myostimulator (Model 7220) was implanted in the subcutaneous area of the canine left thoracic region with its bipolar intramuscular leads implanted around the thoracodorsal nerve and the distal muscle. A total of 6 dogs went through the myostimulator implantation, followed with 8 weeks of continuous stimulation. After completion of the training, the contraction lengths of the trained and untrained LD muscles were measured, and they were 4.25 and 5.5 cm, respectively, while the instantaneous power outputs were 4.24 and 8 W, respectively. Although untrained muscles could provide much higher instantaneous power immediately following the start of the stimulation, it diminished rapidly. On the other hand, the trained muscle showed prolonged fatigue resistance. The thoracolumbar and humeral approaches in attaching the actuator cable did not show a difference with respect to muscle power output. The trained LD muscle can provide sufficient power in the range of 3-4 W to drive a left heart assist device, but its long-term evaluation awaits development of an appropriate muscle-device interface for chronic in vivo application.

In this study, centrifugal pump performance was examined in a mock circulatory loop to derive an automatic pump rotational speed (rpm) control method. The pivot bearing supported sealless centrifugal pump was placed in the left ventricular apex to aorta bypass mode. The pneumatic pulsatile ventricle was used to simulate the natural ventricle. To simulate the suction effect in the ventricle, a collapsible rubber tube was placed in the inflow port of the centrifugal pump in series with the apex of the simulated ventricle. Experimentally, the centrifugal pump speed (rpm) was gradually increased to simulate the suction effect. The pump flow through the centrifugal pump measured by an electromagnetic flowmeter, the aortic pressure, and the motor current were continuously digitized at 100 Hz and stored in a personal computer. The analysis of the cross-spectral density between the pump flow and motor current waveforms revealed that 2 waveforms were highly correlated at the frequency range between 0 and 4 Hz, with the coherence and phase angles being close to 1.0 and 0 degrees, respectively. The fast Fourier transform analysis of the motor current indicated that the second harmonic component of the motor current power density increased with the occurrence of the suction effect in the circuit. The ratio of the fundamental to the second harmonic component decreased less than 1.3 as the scction effect developed in the circuit. It is possible to detect and prevent the suction effect of the centrifugal blood pump in the natural ventricle through analysis of the motor current waveform.