A hospital in the United Kingdom has completed the world's first brain surgery to be carried out with live artificial intelligence assistance, in an operation that removed a tumour threatening the eyesight of a 48-year-old father. The procedure marks the first occasion on which an AI system has analysed a surgical video feed in real time to guide surgeons through one of the body's most densely packed anatomical regions.

The patient, a customer services manager, had been healthy and active until earlier this year, when his condition changed abruptly. He suffered a sudden seizure, followed by persistent fatigue and a gradual narrowing of his peripheral vision. Scans revealed an 11mm tumour on his pituitary gland, a pea-sized organ at the base of the brain that regulates hormones controlling growth, metabolism and reproduction. Though the growth was non-cancerous, its position placed it close to the optic nerves, and its continued growth risked permanent vision loss.

The pituitary gland sits in a narrow space at the skull base, hemmed in by the carotid arteries, which supply blood to the brain, and the optic nerves and chiasm, which carry visual signals to the brain. Surgeons operating in this region have almost no margin for error. Data on standard procedures of this type show that surgeons leave residual tumour tissue behind in a substantial proportion of cases, while injury to a carotid artery, though rare, carries a serious risk of stroke or death. These figures underline why the operation is regarded as among the most technically demanding in neurosurgery.

Surgeons accessed the tumour through the patient's nose, using an endoscope to reach the base of the skull without the need to open the skull itself. As the procedure took place, the AI system analysed the live video feed from the endoscope, tracking the position of surgical instruments and overlaying digital markers onto the footage to indicate zones containing hidden arteries and nerves. The effect has been compared to the facial recognition software used in smartphones, though here the system is trained to identify blood vessels and nerve tissue beneath the surface of human anatomy rather than facial features.

The tool was developed using footage from hundreds of previous operations, building a reference library of surgical experience that exceeds what any individual surgeon is likely to encounter across an entire career. The lead surgeon on the case described the system as functioning like an expert second pair of eyes in the operating theatre, capable of flagging risks that might otherwise go unnoticed amid the pressure of live surgery. Previous AI applications in surgery have tended to operate after the fact, analysing recorded footage once an operation has concluded. This system instead processes information as the surgery unfolds, a distinction that its developers regard as central to its value.

The patient agreed to be the first person to undergo the procedure, saying he wanted to contribute to the advancement of medical research. Reflecting afterwards, he said the experience had left him convinced of the technology's potential to improve safety for future patients, noting that surgeons operating deep inside the brain have no equivalent of road signs to warn them of what lies ahead. He has since returned to his usual activities, with his sight preserved.

Clinicians involved in the project said they expect the technology to be trialled in further procedures involving delicate anatomy, where the margin between a successful operation and a serious complication can be measured in millimetres. Its developers are continuing to expand the system's training data, with the aim of refining its accuracy across a wider range of surgical scenarios.

‍