AI assisted live brain surgery saves UK patient's sight in world first

[Shane_8]

A 48 year old man from Bedfordshire has become the first patient in the world to have a brain tumour removed with the help of live artificial intelligence guidance. Rhys Hibbert underwent the surgery at the National Hospital for Neurology and Neurosurgery in London, part of University College London Hospitals NHS Foundation Trust. The operation successfully removed the tumour and protected his sight, which had been under serious threat beforehand. It is a fairly remarkable milestone for a hospital that has been doing neurosurgery since 1859

What made this different from a normal AI assisted procedure is that the system analysed the live surgical video feed in real time rather than relying purely on scans taken before the operation. The tumour sat on Hibbert's pituitary gland, an area surrounded by blood vessels and nerves controlling vision, where a margin of error of just a millimetre can mean the difference between a clean removal and permanent damage. The AI helped the surgical team identify these risky structures while they worked to remove as much of the tumour as safely possible. Surgeons remained fully in control throughout, with the tool acting purely as an extra layer of visual guidance

Hibbert was diagnosed back in December 2024 after collapsing during a walk and suffering a seizure, which is a frightening way to first learn something is wrong. As his symptoms got worse he chose to go ahead with surgery and volunteered to take part in the research trial using this new tool. He has said since that without patients being willing to join trials like this, medicine has no way to actually progress. That kind of attitude from a patient facing an uncertain procedure is worth acknowledging

The AI tool itself was developed in house at the UCL Hawkes Institute and trained using hundreds of videos from previous pituitary surgeries. This is not the first time AI has entered a UK operating theatre this year either, since surgeons at St Mark's used a tool called Eureka, developed in Japan, during a bowel resection back in June. Between the two cases it looks like UK hospitals are moving fairly quickly to bring these tools out of pure research settings and into live patient care. Whether that pace continues will likely depend a lot on how these early cases hold up over time

This feels like one of the more meaningful uses of AI in medicine right now, given how thin the margins for error already are in this kind of surgery before you even add a computer into the mix. It is easy to be skeptical of AI hype in general but colour coding critical anatomy in real time during a delicate procedure is a pretty concrete, low nonsense application. Curious what other surgical specialties people think could benefit from something like this given how narrow and specialised the current tool clearly is

Ai Helps Perform Brain Surgery On British Man In World First 13577615news.sky.com

[Olivia78]

Training the model on hundreds of past surgery videos rather than general medical imaging data is the detail that stands out to me. It suggests a narrow and specialised tool built for one specific job rather than some broad medical AI system. That kind of focus is probably why it worked well enough to trust with a live patient

[Benzema]

A millimetre of margin for error around the pituitary gland is unsettling to think about, that is an area packed with structures controlling vision and hormone regulation. Having a second set of eyes, even a computerised one, watching for that kind of risk in real time seems like an obvious win once you frame it that way. The surgeon staying fully in control the whole time is the important detail that some headlines are going to gloss over. This was not the AI performing surgery, it was the AI helping a human surgeon see more clearly. Big difference and worth being precise about

[IronWarden12]

Good to see this framed correctly as AI assisting a human surgeon rather than replacing one, that distinction matters a lot for how the public perceives these stories. The actual surgery was still performed by a person making every real decision. AI highlighting risky structures is closer to a very advanced pair of glasses than anything resembling autonomy

[RicFlair_X]

Every ten to twenty pituitary operations a typical UK neurosurgeon performs each year according to some of the coverage, which is not a huge number to build deep individual experience from. A tool trained on hundreds of past surgeries could help standardise outcomes across hospitals that do not see this specific procedure as often

That consistency angle does not get talked about enough in these stories, it is usually framed purely as a technology headline. The actual clinical benefit might be less about any single dramatic case and more about raising the floor for less experienced surgical teams

[MiniElliot]

The part about it being trained purely on past surgery footage rather than some general purpose medical model makes me more comfortable with how it was used here. Purpose built tools for narrow high stakes tasks feel like the right way to introduce AI into medicine rather than trying to make one system do everything. Hopefully this leads to more transparency about validation data as these tools get rolled out more broadly. Patients deserve to know exactly what these systems were trained on before being asked to consent to something like this

[AntiMatterX69]

It is worth remembering this was one patient and one successful case, not a broad clinical trial with statistically meaningful numbers yet. Encouraging news for sure but still very early days for this specific application