The AI will see you now: How robots are replacing surgeons – and achieving good results

In the 1972 science fiction film Silent Running, three spaceship service robots are reprogrammed to perform surgery on a grisly leg injury suffered by the movie’s main character. Metal arms and tools cut away clothing and stitch the bloody wound before, in a surprisingly touching moment, one of the faceless drones gently wraps the injury in bright white bandages.

It is often said that science fiction is on the verge of becoming fact, and nowhere more so than in surgery. As artificial intelligence technology advances at a blistering pace, AI-powered robots are set to perform an increasingly wide range of operations in place of highly skilled physicians, and may sometimes even achieve better results.

For at least some operations, the surgeon’s role could become one of oversight, monitoring the robot and making adjustments to improve accuracy, or intervening if things go wrong. More tasks are likely to be “peeled off and handled by the robot system”, allowing the surgeon to focus on the more critical aspects of a given procedure, according to Dr Jeremy Brown, an assistant professor in the Department of Mechanical Engineering at Johns Hopkins University in Baltimore, US.

“The more we do that, the more systems are going to learn … It’s going to be more and more autonomy with systems as they progress,” Dr Brown said during an online briefing for journalists organised by the university. Among the factors driving the adoption of robotic surgery is an increase in the number of operations as the global population ages.

About 310 million major surgeries are carried out annually, with robots already involved in about four million, such as in preparing joints to receive a replacement hip. Laser eye surgery and radiotherapy are other procedures in which robotics are already involved, while in a research setting robots have carried out gallbladder removal.

“There is an upcoming health care crisis, particularly in surgery. We have fewer surgeons available,” said Dr Axel Krieger, an associate professor of mechanical engineering who researches robotic surgery at Johns Hopkins. “We also have an ageing, growing society, so the caseload is projected to rise more than twofold in the next 10 years. We really need more assistance to keep up with that rising caseload.”

Another factor driving the move to robotics, Dr Krieger said, is the transition towards minimally invasive procedures.

A thin, flexible endoscope tube can be passed either through the mouth, another natural orifice or a single incision into the body to provide images or video, with a surgeon watching on a screen and remotely controlling instruments sent in with the endoscope. Such operations typically involve less blood loss or scarring, and a faster recovery.

Next level

For more than two decades various types of robot systems have been in use, with some made up of an arm equipped with several multi-jointed instruments. Increasingly, AI trained on vast amounts of data from previous operations, these systems could control the instruments or provide guidance to the surgeon.

“These procedures are becoming harder and harder for the surgeons to perform,” Dr Krieger said. “We need more assistance. Robots can play a huge role in enabling that minimally invasive technology.”

Such AI assistance can be helpful in, for example, cochlear implant surgery, where incredible precision is needed. Here, the surgeon is operating as close as 1mm to key facial nerves, damage to which could partially paralyse the patient’s face.

Dr Krieger’s research team has developed a two-armed robot that can, for example, close tubular structures for bowel surgery. The AI that controls the robot improves as it is trained on more data.

“All the architecture needs is the video input and it generates the robot action directly,” Dr Krieger said. “How it learns is by watching human experts do this procedure. So we do different demonstrations of these subtasks and give that to the learning model and it can fully execute these surgical subtasks.”

Suturing is carried out autonomously with a surgeon watching and making adjustments, with the robot producing stitches that are 3mm apart – and more evenly spaced than those by a surgeon.

“We can outperform expert surgeons on some key metrics,” Dr Krieger said. “We have fewer of these hesitancy events when we misplace a needle and have to pull it back out. The goal is to reduce complications, democratise access to everyone for expert surgery and alleviate the shortage of trained surgeons.”

Getting the message across

A key issue with robotic surgery or robot-assisted surgery (RAS) is whether it will be accepted by both the public and medical professionals.

Dr Aseel Takshe, chair of the Public Health Department at the Canadian University Dubai, said research indicated that there were generally “positive perceptions” of RAS among UAE healthcare employees.

“RAS is being adopted in several surgical specialities in the UAE, including general surgery, urology, brain surgery and obstetrics/gynaecology,” she said.

“While there are some reservations – cost and lack of RAS training, limited public awareness of RAS, potential malfunction during the procedure – healthcare professionals in the UAE recognise RAS’s potential benefits for precision, patient outcomes and advancing surgical capabilities.”

Healthcare providers can do much to ensure public acceptance, she said, including by ensuring that patients are well informed and that they understand robot-assisted surgery and its benefits.

“Hospitals offering RAS are often perceived as more advanced or better quality,” she said. “This perception can influence patient choice of healthcare providers.”

In future, robotic surgery will not just be about using robots to carry out existing procedures: computer simulations will allow for new procedures to be developed for AI-powered robots.

Dr Mathias Unberath, an associate professor in Johns Hopkins University's Department of Computer Science, said that a clinical study was not required to trial new types of surgery. Instead, computer simulations could develop new procedures and train the AI.

“The only thing I need is a computer and a little bit of time on the graphics card,” he said. “We can generate millions of highly realistic data samples with the corresponding ground truth directly from human-based models.”

He said that autonomy would become “a ubiquitous part” of surgery in at least the mid or long term, with the role of humans in the operating theatre fundamentally changing to one of supervision.

“It is the beginning of something that is going to be very exciting, especially as AI and machine learning technology is growing stronger by the minute,” he said. “There is a whole lot of reason for optimism and excitement because this technology that we’re developing today is going to have a transformational impact on health care and therefore on the benefits to patients.”