World’s first AI-assisted brain surgery removes tumour:How AI guides surgeons in real time to avoid hidden nerves

For the first time, artificial intelligence has been used live inside a brain operation on a patient, not to perform the surgery, but to help surgeons see and identify critical structures that could not be easily spotted. The world-first procedure was carried out in London on 48-year-old Rhys Hibbert, whose 11mm tumour was growing dangerously close to the nerves controlling his vision and major blood vessels supplying his brain. The surgery was successful, and Hibbert says his eyesight and energy have improved dramatically. What was wrong with the patient? Hibbert, a father of two from Bedfordshire, was fit and healthy until he suddenly collapsed and suffered a seizure while out walking. A scan revealed an 11mm non-cancerous tumour on his pituitary gland, a small organ located at the base of the brain. The pituitary gland produces hormones that help control several important functions, including growth, metabolism and blood pressure. The tumour itself was not cancerous, but its location made it dangerous. It was pressing against the optic nerves, which control vision, and was located close to the carotid arteries, which carry blood to the brain. As the tumour grew, Hibbert’s peripheral vision became narrower. He also began experiencing severe tiredness, dizziness, and problems with balance. Without surgery, his eyesight could have continued to deteriorate, and he could eventually have lost his vision. Why was removing the tumour so difficult? The biggest challenge was the location of the tumour. The pituitary gland sits in a tiny space surrounded by important nerves, blood vessels and other delicate structures. There is very little room for error. Surgeons had to remove as much of the tumour as possible while avoiding the optic nerves and carotid arteries. A mistake could have resulted in severe complications, including blindness, stroke, or even death. And there was another problem: the anatomy inside every person’s head is slightly different. Doctors can study brain scans before surgery, but once the operation begins, they still have to identify and navigate the structures in front of them. This is where the AI system offered something new. How did surgeons reach the tumour? The surgeons did not open Hibbert’s skull. Instead, they used an endoscope, a thin tube with a tiny camera, and passed it through his nose towards the base of the skull. This approach allowed them to reach the tumour without making a large opening in the skull. The endoscope sent a live video of the operating area to a screen. But some of the most important structures were hidden behind bone and tissue, making them difficult to identify. The AI was designed to help with exactly this problem. Also read: Meta agrees to $18 billion deal over child safety lawsuit: Agreement applies to participating US states; case could further impact Indian users

How did the AI assist during surgery? While the surgeons operated, the AI system watched the live video feed from the endoscope. It analysed the footage in real time and identified important structures such as blood vessels and nerves. It also tracked the surgical instruments and marked important areas on the screen, helping doctors understand which parts of the area around the tumour needed to be avoided. Think of it as an extra pair of highly trained eyes inside the operating room. The system works somewhat like facial-recognition technology. Instead of recognising a person’s face, however, it recognises important anatomical structures. The AI could alert surgeons to where critical structures were likely to be, helping them choose safer areas from which to remove the tumour. But there was one crucial rule: the AI was only an assistant. The surgeons remained in complete control and made all decisions during the operation. Prof Hani Marcus, who helped perform the procedure, said: It can act like an expert second pair of eyes. How did the AI learn what to look for? The system was not simply given medical scans and asked to figure everything out. Researchers trained it using hundreds of videos of previous pituitary tumour operations. They carefully marked important structures in the videos, including blood vessels, nerves, surgical instruments and tissue. This gave the AI thousands of examples of what these structures can look like during an actual operation. The researchers say the system has therefore been exposed to a huge variety of surgical situations, potentially more examples than an individual surgeon would encounter in their career. Dr Sophia Bano, the technical lead for the AI system, said: It is designed to help recognise critical anatomy, surgical instruments and tissue interactions in real time. What could the AI actually see? The AI was watching the live surgical video, rather than simply looking at Hibbert’s scans before the operation. This could be especially useful during complicated procedures where important anatomy may be difficult to distinguish. Also read: Haven’t bought a Rakhi gift yet for your sibling?: 7 electronic items under ₹1,000 you can get delivered in minutes

What happened after the surgery? The operation successfully removed the tumour while protecting Hibbert’s vision. He noticed a major difference as soon as he woke up. When I opened my eyes after the operation, it felt like I had got 360-degree vision. His eyesight continued to improve in the weeks after the operation. He also said his energy had returned and the symptoms he had been experiencing before surgery had disappeared. “It’s given me my life back.” Is this the future of AI in surgery? This operation does not mean that AI can now perform brain surgery on its own. In fact, the researchers stress the opposite: the technology is designed to support surgeons, not replace them. The team is now working towards larger trials to test how accurately and safely the system performs across more patients and operations. The long-term aim is even more ambitious: to create an AI assistant that surgeons can consult during difficult procedures, much like having another expert in the operating room. Prof Marcus described the idea as a kind of ‘ChatGPT for surgeons,’ an AI that doctors can turn to for information when they need it, while still having the final say.