What happened in the world’s first real time AI guided brain tumor removal at London’s National Hospital for Neurology and Neurosurgery—including who the 48 year old patient was, the benign 11mm pituitary tumor he had, his symptoms and recovery, how the UCL Hawkes Institute AI an

UCL/UCLH reported the world’s first use of real-time AI assistance during live brain-tumour removal at London’s National Hospital for Neurology and Neurosurgery: the system helped surgeons remove a pituitary tumour while preserving the patient’s vision. The AI was a decision-support tool, not an autonomous surgeon: the human team retained responsibility for interpreting its prompts and making every operative decision. 1

Patient and outcome

• The patient was Rhys Hibbert, 48, a Bedfordshire customer-services manager and community volunteer. 1
• He had an 11 mm benign pituitary tumour. It was discovered after he collapsed on a walk and had a seizure in December 2024; untreated, the tumour was threatening his sight and could ultimately have caused blindness. 1
• The tumour was successfully removed and his vision was protected. The available evidence does not provide fuller clinical follow-up or a detailed account of his longer-term recovery. 1

What the AI did—and why the case was high-risk

• During the endoscopic operation, UCL’s system processed the live surgical video feed rather than relying solely on pre-operative scans. It highlighted critical structures at the skull base, including blood vessels and nerves involved in vision, and was intended to recognise surgical instruments and instrument–tissue interactions. 1
• This anatomy is exceptionally unforgiving: the marble-sized pituitary gland sits tightly among major blood vessels—including the carotid arteries—and optic nerves. A millimetre-scale error can cause catastrophic bleeding, stroke, blindness or death. 1
• Its practical role was to flag hazardous anatomy and areas to avoid, helping the surgeons maximise safe tumour removal; it did not direct, control, or replace the surgeons’ judgement. 1

Team, technology, training, and funding

• Professor Hani Marcus, a UCL Queen Square Institute of Neurology professor and NHNN consultant neurosurgeon, performed the operation with Mr Danyal Khan, a UCL PhD candidate and UCL/UCLH neurosurgical resident leading the work. 1
• The system was developed by a multidisciplinary UCL Hawkes Institute team, with Dr Sophia Bano as technical lead. 1
• It was trained and evaluated on a large, annotated collection—described by Bano as hundreds—of previous endoscopic pituitary-surgery videos, exposing it to a breadth of operative examples that would take an individual surgeon years to encounter. 1
• It ran on NVIDIA’s Clara IGX platform, designed for real-time AI in medical-device settings. 1
• The trial was funded by the National Institute for Health and Care Research and Google, with support from the NIHR UCLH Biomedical Research Centre, Royal College of Surgeons, EPSRC and Wellcome. 1

What comes next

• The programme’s broader ambition is often described as a “ChatGPT for surgeons”: a real-time surgical co-pilot that can understand the operative scene, identify anatomy and actions, and eventually provide timely guidance or feedback. The available evidence supports that direction but does not establish that the system currently has general conversational or autonomous capabilities. 12
• UK Health Innovation Minister James Frith called the case an example of AI enabling previously “unimaginable” care, while stressing that AI requires proper safety safeguards; NIHR’s Mike Lewis said it aligned with the government’s plan to use innovation and technology to improve care. 1
• This was an early clinical-trial case, not proof of benefit across patients. Insufficient evidence is available in the supplied sources to specify the size, design, sites, or timetable of the larger planned clinical trial. 1