Surgeons have performed the world’s first brain operation in which artificial intelligence assisted the team in real time. The system analyzed video from the procedure and guided the doctors around important but hidden blood vessels and nerves, helping them remove as much of the tumor as possible while maintaining safety.

The patient was 48-year-old Ris Hibbert from Bedfordshire, a father of two and a customer service manager. Without the operation, he could have lost his sight. The procedure was carried out by neurosurgeons Professor Hani Marcus and Danial Khan at the National Hospital for Neurology and Neurosurgery, part of University College London Hospital.

Until 18 months earlier, Hibbert had been in good health and walked about 3.2 kilometers every day during his lunch break. During one of those walks, he lost consciousness and had a seizure in the street. A brain scan revealed an 11-millimeter benign tumor growing on the pituitary gland at the base of his skull.

The pituitary gland is a pea-sized organ that releases essential hormones involved in growth, metabolism, blood pressure and body temperature. It lies very close to the carotid arteries, which supply blood to the brain, and to the optic nerves responsible for vision. As Hibbert’s tumor grew, it began pressing on his optic nerves and narrowed his peripheral vision.

An immediate operation was not initially considered necessary, but Hibbert began tripping over objects and experiencing severe tiredness and dizziness. He says the condition had its greatest impact on his emotional and mental health. Although people kindly offered to help, he did not want to become a burden. Hibbert recalls that during his 18 years of marriage he had always got out of bed and made coffee for his wife, and he was determined not to let the tumor stop him.

When surgeons offered him the opportunity to become the first patient to receive the specially developed AI tool during surgery, he accepted without hesitation. “If patients are not willing to take part in research, how will doctors ever learn and how will medicine advance?” he said.

The operation involved inserting an endoscope—a thin camera mounted on a rod—through the nose to reach the base of the skull. The tumor was close to critical structures but hidden behind layers of bone and healthy membrane. Finding the safest route for its removal was difficult, particularly because each person’s anatomy is different.

Depending on the case, the chance that a tumor will not be completely removed is between 25% and 50%, while the risk of damaging a major blood vessel ranges from 0.5% to 2%. During the operation, the AI system analyzed the video on a second screen, tracked the surgical instruments and marked areas where blood vessels and nerves were most likely to be located. It then indicated the places where removing the tumor was safest.

The technology resembles facial-recognition systems used in phones, but instead of identifying faces it recognizes important anatomical structures that are often hidden. Researchers trained and tested it using hundreds of videos of pituitary-tumor operations. Specialists carefully outlined blood vessels and nerves in each recording, allowing the system to collect data and begin recognizing the most important structures.

An average British surgeon performs between 10 and 20 similar operations a year and generally relies on scans made before surgery to understand an individual patient’s anatomy. Professor Marcus says the system has been trained on more operations than most surgeons see or perform in their entire careers and can serve as a “second expert pair of eyes.” Unlike other experimental AI tools, he added, this system operates in real time.

The team stresses that the artificial intelligence only assists the doctors, who retain full control of the operation. Their long-term goal is to develop what Professor Marcus describes as a kind of “ChatGPT for surgeons”—another expert in the operating room whom doctors could consult for advice, or ignore if they disagreed with it.

The project is funded by the National Institute for Health and Medical Research and Google. The team tested the technology in laboratory conditions for years. After the first operation on a real patient, the researchers said they were pleased with the initial results and were already working on a larger clinical trial.

Eight weeks after the procedure, Hibbert said his vision was improving every day. When he opened his eyes after surgery, he felt as if he could see in 360 degrees and said he had not seen so clearly for a year. His energy also returned, while the side effects he had experienced before the operation disappeared. “It gave me my life back,” he said.

Britain’s health innovation minister, James Frith, said he was pleased that Hibbert had undergone a “life-changing operation” as part of government-funded research. He added that artificial intelligence required appropriate safeguards and that safety would always come first, while the opportunities offered by the technology should also be pursued.