“We tried other software, but it was harder to create models and systems,” says research assistant Mert Yiğit, a recent biomedical engineering graduate specializing in turbine design and computational fluid dynamics. “MATLAB is easy to learn, and you can create anything that comes to mind.”

The lab’s mechanical circulatory support system includes an implantable heart pump powered by batteries, a smart controller, and a wireless portable patient unit for vitals monitoring. Students made nearly all the hardware for their prototypes on-site.

While considering the path to clinical trials, the young YTU team came up with an approach to minimize animal testing for their LVAD and other heart devices. They built a hybrid pneumo-hydraulic mock circuit in MATLAB that enables rigorous testing in realistic cardiovascular conditions. The students optimized and accelerated computations using GPU Coder™ and a powerful NVIDIA^® workstation. 

“The students are the dynamo behind this whole project,” Kadıpaşaoğlu says. Their research was published in the International Journal of Robust and Nonlinear Control.

Hands-On Experience

“Coming from Turkey, we have a hard time financing these devices,” Kadıpaşaoğlu observes. “And Turkish surgeons haven’t gone through a research phase with device engineers, so they’re inexperienced. They would implant pumps in patients after training for a day or two, sometimes with disastrous results.”

He brought his knowledge and experience back home to teach enterprising YTU students about LVAD development, foster academic ties to the medical sector, and lay the groundwork for manufacturing affordable heart pumps in Turkey. Kadıpaşaoğlu serves as a mentor to students on biomedical, electrical, and control engineering career paths.

“I’m trying to help them secure training in hospitals and surgical theaters so that they can have hands-on experience with blood, open chests, and beating hearts,” he says. 

Kadıpaşaoğlu encourages self-directed research. Mini portraits of influential scientists such as Sir Isaac Newton and Joseph-Louis Lagrange line the lab walls for inspiration. There are around 20 students in the group, including two who graduated summa cum laude: Derya Sahin and Ahmed Alhajyounis. The students come up with projects, write grants, and coauthor journal articles.

Yiğit displayed the team’s latest LVAD prototype, a rigid 7.2-centimeter (2.8-inch)-long cylindrical turbine that the group crafted from biocompatible titanium alloy.

Each motor’s moving sections are made from a soft magnetic composite and a plastic template that fits on top, holding mini magnets. They also used the magnetic composite for the toothed stationary core, wrapping thin copper wire around it to produce a magnetic field that interacts with the moving part.

Early on, the students waited an entire year for an outside group to produce a single motor. “When our team decided to build our motors, we did it in one month,” Yiğit explains.

Realistic Cardiovascular Device Testing

The lab used Simulink and MATLAB to construct the complex cardiovascular model of this engineered system, where several systems function in series with one another. Adjusting a single parameter in one system influences the other parameters in tandem, Kadıpaşaoğlu pointed out.

“Simulink makes it easy to build a cardiovascular system simulator, but the most difficult part is adjusting for the desired patient-specific outcomes,” he says.

Yiğit unveiled the latest physical hybrid mock circuit setup in the lab, with their most recent LVAD prototype visible inside a sleek water-filled chamber connected to hydraulics. He explained that creating controllers for the pump flows and pressures inside these physical components was extremely difficult, as the pneumatic systems are nonlinear.

“Pressure accumulates inside, so if you enter the wrong inputs, water could spray all over the place,” Yiğit cautions.

To find the control coefficients, they created a digital twin of their hybrid mock circuit in MATLAB. The group leveraged Q-learning, a reinforcement-based approach, with Statistics and Machine Learning Toolbox™ to identify the control coefficients for their cardiovascular system. PCL members developed a machine learning algorithm that automatically adjusted the controllers. Now, if someone enters the wrong input, there’s no liquid explosion.

The group tackled other challenges. Cardiovascular modeling data resembles a bowl of spaghetti. Flow imaging, called particle image velocimetry (PIV), enables researchers to visualize blood flow patterns and velocities. The team worked on image processing and reconstruction in MATLAB, but discovered that simulating fluid dynamics, blood flow, and cardiovascular system scenarios was computationally demanding.

“A single simulation used to require one minute to process all the information. Using GPU Coder with the Jetson, it now takes only 10 seconds,” Yiğit says.

Through this setup, the students can easily test their cardiovascular model under varying conditions while making frequent adjustments to the physical parameters. Additionally, they utilized MATLAB Coder™ to convert their MATLAB algorithms into C and C++ code for deployment on a dSPACE^® real-time processing system, allowing them to operate the hybrid mock circuit. They successfully validated their heart pump prototype’s dual motor system in the hybrid mock circuit by simulating a patient’s physiological conditions with just one operational motor.

PCL members have a new effort underway that leverages machine learning to determine the optimal blood flow rate for the LVAD prototype. And the entire team is working to reduce the size of the heart pump by 20% while improving its efficiency. They also seek to make their hybrid mock circuit more compact and bring it to market.

The group plans to eventually replace the dSPACE real-time processing system with the Jetson TX2 board to greatly increase the performance of the real-time simulation. The group’s use of GPU Coder in the simulation phase will help ease the migration.

“We’re continually asking ourselves whether a technology is something we have to buy or something we could create ourselves,” Yiğit says. “If we can create it on our own, we develop a project around it.”

Kadıpaşaoğlu reflected that he’s more than met his goals since launching the lab at Yildiz Technical University. “We are starting to get our name on the map,” he says. “People are coming to us for interviews. The laboratory has become a place where other universities send their undergraduate students to train.”