{"id":216,"date":"2025-07-09T17:28:13","date_gmt":"2025-07-10T00:28:13","guid":{"rendered":"https:\/\/scienceblog.com\/neuroedge\/?p=216"},"modified":"2025-07-09T17:28:13","modified_gmt":"2025-07-10T00:28:13","slug":"robot-surgeons-learn-like-residents-and-just-performed-first-autonomous-surgery","status":"publish","type":"post","link":"https:\/\/scienceblog.com\/neuroedge\/2025\/07\/09\/robot-surgeons-learn-like-residents-and-just-performed-first-autonomous-surgery\/","title":{"rendered":"Robot Surgeons Learn Like Residents\u2014And Just Performed First Autonomous Surgery"},"content":{"rendered":"

Three surgical robots trained on video demonstrations have achieved what many thought was years away: performing complex procedures with the precision of expert surgeons and the adaptability to handle the unexpected twists that define real medical emergencies. The systems mark a watershed moment where artificial intelligence moves from simple surgical assistance to genuine autonomous decision-making in the operating room.<\/p>\n

The most striking demonstration came from Johns Hopkins University<\/a>, where a robot completed an entire phase of gallbladder removal surgery without human intervention. But this wasn’t just mechanical precision\u2014the robot responded to voice commands, learned from mistakes in real-time, and adapted when researchers deliberately changed conditions mid-procedure.<\/p>\n

Beyond Programming: Robots That Actually Understand Surgery<\/h2>\n

What separates these systems from earlier surgical robots isn’t just capability\u2014it’s comprehension. Traditional surgical robots follow rigid, predetermined paths like “teaching a robot to drive along a carefully mapped route,” explains Johns Hopkins medical roboticist Axel Krieger. His new system “is like teaching a robot to navigate any road, in any condition, responding intelligently to whatever it encounters.”<\/p>\n

The robot, called Surgical Robot Transformer-Hierarchy (SRT-H), learned by watching videos of Johns Hopkins surgeons performing gallbladder procedures on pig cadavers. After analyzing 17 hours of surgical footage across 34 different specimens, the system performed a complex 17-step procedure with 100% accuracy on eight different gallbladders\u2014each with unique anatomical features.<\/p>\n

Built using the same machine learning architecture that powers ChatGPT, SRT-H can respond to spoken commands like “grab the gallbladder head” and corrections such as “move the left arm a bit to the left.” More remarkably, it learns from this feedback during surgery.<\/p>\n

The Needle Navigation Challenge<\/h2>\n

Meanwhile, researchers at the University of North Carolina at Chapel Hill<\/a> are pioneering what they call “AI guidance” for medical needle procedures\u2014those critical moments when physicians must thread instruments through mazes of blood vessels and airways to reach targets as small as a pea.<\/p>\n

Traditional image guidance has helped doctors visualize anatomy since X-rays were discovered in the late 1800s. But AI guidance goes further, automatically analyzing images, identifying targets and obstacles, computing safe trajectories, and even steering robotic needles autonomously around sensitive tissues.<\/p>\n

The team demonstrated their system navigating needles to clinically-relevant targets in living lung tissue with better accuracy than physicians using traditional tools. The technology defines four levels of AI involvement:<\/p>\n