Design of a non-beating electromechanical heart

Although artificial hearts have successfully prolonged the lives of some heart patients, several design flaws have long limited their use. Currently, these electromechanical hearts rely on positive displacement blood pumps, which are too large. They are so large that even the latest self-operating hearts cannot easily fit into a small chest cavity. Positive displacement blood pumps also have more moving, cyclically loaded parts than engineers would like to see for stability reasons. A new electromechanical heart being developed at the Texas Heart Institute will use a different blood pump and different precision control algorithms to address both of these issues.

Instead of using a positive displacement blood pump to mimic the beating of a natural heart, the Texas Heart Institute's Total Artificial Heart (TAH) design uses a pair of continuous-flow axial-flow pumps. A pulmonary circulation pump delivers deoxygenated blood to the lungs and returns oxygenated blood to the heart, while a systemic circulation pump delivers oxygenated blood from the heart to the body and returns deoxygenated blood to the heart. Using a controller, the two blood pumps can work in tandem and adjust the output of one blood pump based on changes in one blood pump and physiological demand.

According to Steve Parnis, associate director of technology at the Texas Heart Institute's Heart Support Center, the two continuous-flow axial pumps are actually repurposed DeBakey ventricular assist devices from MicroMed Cardiovascular. Typically, a ventricular assist device (VAD) does what its name implies, assisting the natural heart in pumping blood. "But here, these two VADs completely replace the natural ventricles," Parnis said.

This innovative idea has been around for several years. In 2006, Dr. Bud Frazier, director of research and chief of heart and lung transplantation at the Texas Heart Institute, published several papers on the continuous-flow total artificial heart. In 2008, his idea moved one step closer to clinical practice when the National Institutes of Health awarded the Texas Heart Institute $2.8 million to fund the design of the continuous-flow heart.

In a total artificial heart application, VADs offer several advantages. For one, they are about the size of a C-cell battery, whereas a self-contained pulsatile blood pump is a two-pound block of titanium and plastic. "VADs will fit most patients, whereas only a minority of patients are suitable for today's pulsatile blood pumps," Parnis said.

Second, VADs have a proven clinical track record. About 500 DeBakey VADs are now in use, according to Bob Benkowski, MicroMed’s chief operating officer and one of the engineers who helped develop the original DeBakey VAD model. “They’ve been working for eight years,” he said. And he attributes that stability in part to the simplicity of MicroMed’s axial-flow pumps, whose only moving part, the impeller, is driven directly by an electric winding.

Parnis argues that even the most modern positive flow pumps have a lifespan of only two years; the pulsation of a positive flow pump creates cyclic loading conditions that do not occur with an axial flow pump. He adds that a continuous flow pump would probably require less power and cost less than a more complex pulsatile flow pump. If two continuous flow ventricular assist devices would make such a great total artificial heart, why haven't they been used so far? Because they would still require a lot of controller engineering to make the leap from natural heart assist to total artificial heart replacement.

A new total artificial heart concept uses a pair of small axial flow pumps like the one above to move blood through the body's systemic and pulmonary circulation. One ventricular assist device is used to move blood throughout the body, while the other is used to move blood to and from the lungs. Typically, these blood pumps are used as ventricular assist devices to assist the natural heart.

That's where Matthew Franchek and Ralph Metcalfe, both mechanical engineers with Ph.D.s and professors at the University of Houston's Cullen College of Engineering, come in. As part of a grant from the National Institutes of Health, they are working together on a feedback controller that would allow two ventricular assist devices to work together as a total artificial heart. The university researchers have helped develop similar automatic regulating control systems for automotive applications, and most of them are now working on an automatic governor for diesel engines for Cummins Engine Company.

[page]

In some ways, Franchek and Metcalfe were the first to start controller development because they were using mature VAD technology. MicroMed's VADs already had their own controllers. Benkowski described these controllers as feedback controllers that use ultrasonic sensors to measure actual flow, compare actual flow to the desired flow output, and then generate an appropriate pulse width modulated control signal to adjust the impeller speed.

However, the two engineering professors still had their work cut out for them. Typically, ventricular assist devices work independently as auxiliary support for the residual natural heart. In a total artificial heart, they must work closely together to mimic the balanced flow of the left and right ventricles of a natural heart. Franchek said: "Pairing two blood flow pumps together must face a complex and variable control problem. The load state and flow output of each blood flow pump will affect the load state and flow output of the other blood flow pump."

The artificial heart controller must also closely link these related flow and load states, including inlet pressure and outflow resistance, to the changing needs of the human body. Everyday activities such as standing or walking can change flow and load states, Franchek said. Cardiovascular events such as vascular restriction, hypertension or changes in blood viscosity and inherent physiological differences between different patients can also affect flow and load states. "The challenge is that we have to maintain a steady-state cardiac output regardless of the reason for the fluctuation in physiological state," Franchek said.

Axial flow pumps are here to meet this need. With the help of feedback controllers, they can be adjusted automatically because their flow output is very sensitive to both inlet pressure and outflow resistance. Benkowski said that the geometry of the ventricular assist device blood flow pump impeller and the structure of the blood flow tube are designed to match the optimal flow pressure behavior for total artificial heart applications. "We can change the pressure sensitivity of these blood flow pumps to make the control algorithm work a little easier," he said.

An integral controller allows two axial flow pumps to work together and automatically adjust their output flow rates based on changes in physiological state. The controller algorithm incorporates a mathematical model of the dynamic behavior of the circulatory system. The sensed value is subtracted from the ideal value to generate an error signal, which is amplified by the controller.

At the same time, the control algorithms will be implemented based on an analog integral controller that measures the actual output flow and compares it to the ideal flow value to adjust the voltage accordingly. Franchek and Metcalfe chose a seemingly simple integral control strategy for this application because it performs very well in maintaining the state of the system in a system whose dynamic behavior is well understood and has cooperative transient characteristics. Since physiological state has an impact on the state of the blood pump, understanding this dynamic behavior is not an easy task. Moreover, most of the development work of the controller funded by the National Institutes of Health is related to creating a lumped parameter mathematical model of the human circulatory system. According to Franchek, this model will eventually be incorporated into the control algorithm of the total artificial heart.

Franchek says he expects the first pass of the total artificial heart's control algorithms to be this summer. "Right now, we're just starting to design and build these controllers," he says. And there are still some important decisions to be made about how the blood pumps will operate. For example, the researchers need to decide whether to have one blood pump operate in a semi-pulsatile state or both. If necessary, Franchek says, the pump motors can be easily "turned on and off" to simulate the pulsating behavior of a natural heart. Other development work includes adding the possibility of monitoring blood viscosity to the system. "We believe that from the flow measurement and the voltage signal, we can effectively infer the viscosity of the blood," Franchek says.

Now, he and Metcalfe are using a variety of simulation tools to carry out the development work. MATLAB and Simulink software are being used to develop mathematical models. They are also simulating the final control algorithm and prototyping the controller hardware in DSPACE, a set of development tools for mechatronic systems. Franchek said he hopes to have the first version of the controller available sometime this summer.