A Johns Hopkins University robot has made history by performing the first realistic laparoscopic cholecystectomy (gallbladder removal) on a lifelike human model without human intervention.
Named Hierarchical Surgical Robot Transformer (SRT-H), the machine executed 17 intricate surgical tasks with 100% accuracy. It adapts dynamically to its environment, responds to voice cues from a surgical team, and makes independent decisions in real time.
Gallstone Disease in the U.S.
According to a 2024 report, gallbladder disease has led to 2.2 million ambulatory care and 1.2 million emergency department visits, 625,000 hospital discharges, and around 2,000 deaths between 2006 and 2019.
Moreover, during the assessed period, U.S. hospitals performed around 605,000 ambulatory laparoscopic, 280,000 inpatient laparoscopic, and 49,000 inpatient open gallbladder surgeries.
Think Like a Surgeon
With laparoscopic cholecystectomies becoming the norm, medical roboticist and SRT-H’s creator, Axel Krieger, believes SRT-H could bring us closer to clinically viable autonomous surgery and improve patient outcomes.
SRT-H, which has a ChatGPT-inspired machine learning brain, studied the university’s “pig cadaver” cholecystectomy videos. These videos were paired with plain-language captions such as “clip the cystic artery” or “expose Calot’s triangle,” allowing the robot to build its own surgical “playbook” that maps objectives to physical movements.
So far, SRT-H has adapted flawlessly to variations in anatomy, identifying arteries, applying clips, and even cutting tissue based on real-time camera input and verbal commands from observing surgeons without committing a single error.
The Road Ahead
SRT-H will have to undergo rigorous safety testing on cadavers and live animals, especially under emergency room-like conditions — think fogged cameras, suction delays, and electrocautery sparks — before it can be tested on live human patients.
To this end, the university’s engineers are developing a “watchdog” module that could instantly pause an SRT-H procedure and alert a human surgeon if it detects errors or sensor drifts.
Image credit: Juo-Tung Chen/Johns Hopkins University