Self-Navigating Catheter Designed for Heart Surgery Tested in Pigs

The robotic catheter can guide its own movements within the heart of a live mammal to the site of a leaky valve replacement.

Apr 24, 2019
Ruth Williams

ABOVE: The robotic cardiac catheter used in the study.

By gently feeling its way around the walls of a beating pig heart, a self-guiding catheter can find a leaky valve and position itself for a human operator to plug the hole, according to a report in Science Robotics today (April 24). The system’s developers suggest that automating catheter navigation could free surgeons from having to perform this challenging yet routine process, allowing them to focus on the most critical aspects of the surgery.

“Building consistency and reproducibility into [surgical] procedures generally improves [patient] outcomes,” heart surgeon and researcher Michael Reardon of Houston Methodist Hospital who was not involved with the research writes in an email to The Scientist. “Robotic control of the catheter delivery systems that we use for endovascular procedures can help us obtain this goal . . . [and] the work by Fagogenis and colleagues provides us with a good start.”

The system “is particularly clever and has real utility,” adds heart surgeon and researcher, Marco Zenati of Harvard Medical School who also did not participate in the study.

Minimizing the invasiveness of procedures is a desirable goal for all types of surgery. By reducing trauma, recovery time is faster, infection risk lower, and cosmetically undesirable scarring is minimized. In the case of heart surgery, replacing open-heart procedures with catheter-mediated ones can also obviate the use of heart and lung bypass machines and eliminate the associated risks, such as arrhythmia and stroke.

Heart surgeons and engineers have produced an array of keyhole surgical techniques, including aortic valve replacements and perivalvular leak closures, both of which are performed by feeding a catheter through either a tiny hole in the chest or the femoral artery into the heart. However, such procedures generally “require exposure to radiation both to the patient and to the providers,” says Zenati, explaining that lengthy imaging with fluoroscopy (a continuous X-ray) is often needed to navigate the catheter into position. A catheter with a tiny ultrasound probe at the tip is sometimes also used but the image quality is poor.

As a possible solution to these imaging shortcomings, says bioengineer and lead researcher Pierre Dupont of Boston Children’s Hospital and Harvard Medical, “our idea was to look at using endoscopy inside the heart together with robotics.” His team chose perivalvular leak closure—where a leak around an implanted replacement valve is plugged—as a proof-of-principle procedure.

Endoscopy is normally performed in air- or clear fluid-filled body cavities, says Dupont, “but if you’re in a blood-filled heart, all you see is red.” That is, “until you press your endoscopic instrument against the tissue and the blood is displaced.”

The team used such visual cues to train its robotic catheter, via machine learning algorithms, to recognize when it was either in contact with the heart wall, surrounded only by blood (away from the wall), or when it had reached the valve (identified by the presence of visible stitches).

The catheter itself was a concentric tube robot—essentially, a flexible metal rod that can maneuver in 3D—with a camera housed in its tip together with an extendable rod for delivering the plug to fix the leak.

In pigs with prosthetic aortic valves that had been purposefully implanted so as to leak, the robotic catheter could navigate from the heart’s apex (the entry point) to the leak almost as quickly and accurately as a human-operated catheter. Once in position, the operator took control, via a joystick, to embed the plug. “The robot does the less critical part of navigating to the leak and then the clinician does the more mission-critical part of deploying the occluder,” says Dupont. He likens surgeons in this context to fighter pilots who could fly to a location on autopilot but then take controls for reconnaissance, battle, or some other essential task.

The technology is not ready for clinical use yet, says Zenati. For one thing, the sutures in the artificial valve were not typical, he says. The team used “a custom valve that had reference points that could be recognized by the system, so in order to apply this in the clinic you’d have to have similar reference points within valves that are implanted.” Nevertheless, the work is a proof-of-principle for autonomous robotic catheter navigation. “The lessons learned . . . could be applied to other [heart] problems” and entry routes, such as the femoral artery.

Indeed, adds Howie Choset, a biorobotics researcher at Carnegie Mellon University who was not involved with the project, “while this [system] was prescribed for heart procedures, I do believe the core idea of using the anatomic walls to navigate is key and can apply to other areas [of the body].”

G. Fagogenis et al., “Autonomous robotic intracardiac catheter navigation using haptic vision,” Sci. Robot, 4:eaaw1977, 2019.

May 2019

AI Tackles Biology

How machine learning will revolutionize science and medicine.