Communications of the ACM
Snaking Its Way into Your Heart
Boston Children's Hospital bioengineers have developed catheter guided by artificial intelligence that can self-navigate inside a simulated beating heart.
Credit: Randal McKenzie
Bioengineers at Boston Children's Hospital have developed a catheter driven by artificial intelligence (AI) that can self-navigate inside a simulated beating heart.
The proof-of-concept research offers the eventual promise of a significant advance in robotic surgery, given that the catheter is completely autonomous.
"Robotics has been and still is investigated for the navigation of instruments such as guide wires and catheters," says Sylvain Martel, director of the NanoRobotics Laboratory of Polytechnique Montréal, a leading Canadian engineering educational and research institutions. "But what differentiates this work is the level of integration of AI to achieve full autonomy."
Douglas L. Jones, a professor of physiology, pharmacology, and medicine in the Schulick School of Medicine & Dentistry of Canada's Western University agrees: "The novelty here is the addition of the AI learning, which does not need external interpretation of haptics registration to accomplish the positioning."
The snake-like catheter works by feeling its way through the heart with the help of an optical touch sensor at its head, which features an LED spotlight coupled with an endoscopic camera. As it makes its journey through the beating organ, the optical touch sensor continually samples its environment to get a sense of its location. That data is continually fed to the system's AI software, which signals the optical touch sensor if the catheter is on the right track, or if it needs to make a minute change in direction to get back on course.
The AI software is pre-programmed with an anatomical map of the specific heart through which the catheter is moving, according to Georgios Fagogenis, a member of the research team and a postdoctoral research fellow specializing in autonomous surgical robots at Boston Children's Hospital. "It knows roughly where it is starting in the map. and it is told what its desired location is," Fagogenis says.
Signals from the optical touch sensor enable the system's AI to discern if the catheter's optical touch sensor is in contact with blood, the heart wall, or a heart valve. Those additional insights further help the catheter to navigate properly and ensure that it moves closer to its destination, rather than ripping a hole in the heart.
That's no easy trick, according to Thomas W. Secord, an assistant professor of mechanical engineering at the University of St. Thomas in Minnesota, who describes the tool's tissue identification capabilities as "impressive."
The researchers' decision to incorporate AI into their system also offers another major bonus: with each use, the AI catheter learns better how to navigate its way through the heart.
In their proof of concept test, the researchers programmed the AI catheter to move itself to a tiny leak located on the side of an artificial valve lodged in the model heart's aorta. The AI catheter pulled off the journey to the leak without a hitch, and a human surgeon was able to take over the operation of the catheter to simulate the "plugging" of the leak.
While Fagogenis says the research team currently has no firm plans to commercialize the AI catheter, other researchers are eager to see a commercialized product, since the tech would free up surgeons to focus on more complex aspects of surgery.
With AI handling positioning of a catheter inside the heart, clinicians could focus more on "planning and decision-making, while robotic technology will enhance the accuracy and repeatability of the operations and decrease the procedure time and hospital stay," says Islam S. M. Khalil, an assistant professor of biomechanical engineering at the University of Twente in the Netherlands.
Ellen Roche, W.M. Keck Career Development Professor in Biomedical Engineering at the Massachusetts Institute of Technology, said that "if carefully implemented into the clinic, this could reduce procedural times and interventionalist fatigue."
The researchers say a commercialized AI-driven catheter could be used to do routine heart-mapping for surgeons; it also could be remotely manipulated to enable highly experienced surgeons to perform particularly difficult surgeries in parts of the world where world-class surgeons are not typically available.
Unfortunately, there are several hurdles to clear prior to commercialization. First, explains Martel, "As the level of autonomy in medical robotic systems increases, the regulatory challenges would most likely increase as well. Therefore, extensive demonstration cycles addressing the safety issues and the advantages over existing practices will be expected before commercialization occurs."
There's also the eek! factor: some patients could be initially repulsed by the prospect of a serpent-like apparatus weaving its way through their hearts, driven by a 'mind' of its own. While Martel concedes such reactions are a concern, he predicts "reluctance from human patients will dissipate as confidence in highly autonomous medical robotic systems will increase from experiencing other autonomous systems, such as self-driving cars."
Ultimately, "It is a question of time for the technologies to become more mature and reliable, and for humans to adapt," Martel says. "But commercialization will certainly happen."
In the long term, the research team foresees a time when AI-driven surgical tools will be in use around the world, and will be networked to enable them to pool their ever-growing, shared expertise with each new surgery completed, according to says Pierre Dupont, the senior investigator on the study and chief of pediatric cardiac bioengineering at Boston Children's Hospital.
By using such AI-enhanced tools, Dupont says, "Every clinician in the world would be operating at a level of skill and experience equivalent to the best in their field. This has always been the promise of medical robots. Autonomy may be what gets us there."
Joe Dysart is an Internet speaker and business consultant based in Manhattan, NY, USA.
No entries found