Surgeons in London remove brain tumor and preserve patient's vision using real-time AI guidance

First clinical use in London

In May 2026, neurosurgeons at the National Hospital for Neurology and Neurosurgery (NHNN), part of University College London Hospitals NHS Foundation Trust, completed the first brain surgery guided by real-time artificial intelligence. The patient, 48-year-old customer service manager Rhys Hibbert from Bedfordshire, faced progressive vision loss from a benign tumor located on his pituitary gland. Surgeons made details of the intervention public on 27 August 2026 following several weeks of post-operative recovery. The procedure allowed the medical team to safely remove the growth while preserving Hibbert's eyesight. During the intervention, the software provided live anatomical guidance directly from surgical video feeds rather than relying solely on pre-operative scans.

From a patient safety perspective, I can clearly see the benefit of this technology. There are no road signs inside our head.

Case progression and diagnosis

Hibbert was in good health until December 2024, when he collapsed and suffered a seizure during a routine two-mile walk during his lunch break. Subsequent diagnostic scans revealed an 11-millimeter tumor growing on his pituitary gland, an organ situated at the skull base that produces hormones regulating growth, metabolism, body temperature, and blood pressure. Over the following 18 months, the growth began pressing against his optic nerves and altered his hormone levels. Hibbert experienced severe fatigue, dizziness, and a progressive narrowing of his peripheral vision, which forced him to use walking sticks after repeatedly stumbling. When clinicians at UCL proposed testing the experimental navigation software during surgery, Hibbert agreed to participate in the clinical trial.

Real-time computer vision in the operating theater

The surgical team, led by consultant neurosurgeon Professor Hani Marcus and surgical registrar Danyal Khan, accessed the tumor by inserting an endoscope through Hibbert's nose to reach the base of the skull. The pituitary gland lies in close proximity to the optic nerves and the carotid arteries, which supply blood directly to the brain. In this tight anatomical space, an error of a single millimeter risks stroke, severe haemorrhage, permanent blindness, or death. Standard procedures carry an estimated 25% to 50% probability of incomplete resection and a 0.5% to 2% risk of critical vascular damage. The AI system analyzed live video from the endoscope, tracking surgical instruments and highlighting safe corridors and critical hidden structures on a secondary monitor.

The system was trained on more operations than most surgeons see or perform in their entire careers.

Development and post-operative recovery

The navigation software was developed by a multidisciplinary team at the UCL Hawkes Institute and funded by the National Institute for Health and Care Research. Technical lead Sophia Bano explained that researchers trained the algorithm on hundreds of recorded neurosurgeries, enabling it to recognize anatomical structures using principles similar to facial recognition technology. Marcus emphasized that the software acted as an expert second pair of eyes, while the surgical team retained manual control of all instruments throughout the procedure. Hibbert reported an immediate improvement in his visual clarity after waking, walked independently within one week, and has since returned to work. UK Health Innovation Minister James Frith stated that the deployment illustrates the capabilities of medical AI while noting the continued necessity of clinical safeguards.