While years of meticulous study and refined dexterity are typically required for humans to master surgery, robots could potentially learn the skill more rapidly thanks to advancements in artificial intelligence.
Researchers at Johns Hopkins University and Stanford University have successfully trained a robotic surgical system to perform complex surgical tasks with the same level of proficiency as human surgeons, solely through instruction via videos of actual operations.
The team utilized a framework to guide their analysis. The robotic system, typically controlled remotely by a surgeon, features articulated arms that precision-manipulate instruments for tasks such as delicate dissection, efficient suction, and precise vessel sealing. Techniques such as these grant surgeons significantly enhanced management, precision, and a more comprehensive assessment of patients during operations. The latest model lacks essential components, including equipment, sterilization tools, and guidance.
Intuitive
By employing a machine learning approach called robotic surgery, the team trained a da Vinci Surgical System to perform three tasks intrinsic to surgical procedures autonomously: manipulating a needle, lifting bodily tissue, and suturing. Have a look.
Surgical Robotic Transformer Demo
The surgical system, capable of executing tasks independently of human oversight, also demonstrated the ability to correct its own mistakes. “When a record drops, it will consistently pick itself back up and continue playing.” Axel Krieger, an assistant professor at Johns Hopkins University, noted that “this is not something I taught it to do,” in reference to the AI system’s unexpected behavior, which was highlighted during a presentation at the recent Convention on Robotic Studies.
Researchers successfully trained an AI model by integrating imitation learning with the machine learning architecture typically used in construction of popular chatbots, such as ChatGPT. Notwithstanding their design for processing text, this model generates kinematics – a language of motion description employing numerical and algebraic components – to guide the surgical system’s arms directly?
Intuitive
The mannequin was trained using numerous movies recorded from wrist cameras installed on Da Vinci robots during surgeries.
The team hypothesizes that their AI-powered mannequin can design a robotic system capable of efficiently executing various surgical procedures with greater speed and ease than traditional methods, which require laborious manual coding of each step to control the robot’s movements.
Johns Hopkins University and Stanford University
By building on Krieger’s insights, the potential exists to bring fully autonomous surgery within reach sooner than previously thought possible. “What’s innovative here is that we can collect imitation data on various processes in just a few hours, allowing us to teach a robot what it needs to learn within two days,” he said. “It enables accelerated progress towards autonomy while reducing medical errors and achieving higher accuracy in surgical procedures.”
That might just be one of the most significant advancements in robotic-assisted surgery globally in recent years. There exist specialized robotic units capable of handling intricate tasks and procedures. Despite these limitations, their abilities typically confine themselves to specific stages of the surgical processes in which they participate.
Coding every step of a robotic system from scratch could potentially lead to slowed performance. “Surgical simulations can require years of dedication and effort, as developers strive to accurately model the intricate process of suturing.” “That’s just one type of surgical procedure requiring suturing.”
Prior to his work on robotic surgery, Krieger had developed an innovative approach to automating surgical tasks. At Johns Hopkins University (JHU) in 2022, his team of researchers designed and developed the Scalable Toolkit for Assessing Risk (STAR). Without human oversight, a robotic system seamlessly stitched together the edges of a pig’s intestine, guided by a novel, structure-based three-dimensional endoscope and a machine learning-driven monitoring algorithm.
Researchers at Johns Hopkins University (JHU) are employing their innovative imitation learning approach to train a robotic system capable of performing a complete surgical procedure. While robotic surgery may still be a distant prospect, advancements in this field can already bring about significant benefits by enhancing the safety and accessibility of complex treatments worldwide.
Supply:
