London neurosurgeons perform world's first AI-assisted brain tumour removal, saving patient's sight

The landmark procedure took place in May at the National Hospital for Neurology and Neurosurgery (NHNN), part of University College London Hospitals NHS Foundation Trust. Details were kept confidential until Thursday while the patient recovered.

Surgeons used an AI system that provided real-time analysis of camera footage during the operation, identifying critical anatomy in the brain that needed to be avoided to prevent damage to the patient's sight. The AI assisted in distinguishing tumour tissue from healthy brain structures, enabling a more precise and safer removal.

Rhys Hibbert, the 48-year-old patient, retained his sight following the operation. Health officials described the outcome as a success and a significant step forward for neurosurgery.

The procedure is thought to be the first of its kind globally, where AI was used intraoperatively to guide the surgeon's decisions based on live video feeds. While AI has been used in pre-operative planning and diagnostic imaging, this case represents a direct, real-time application during the critical phase of tumour resection.

The NHNN, a leading centre for neurology and neurosurgery, has been exploring AI-assisted techniques for several years. This surgery builds on earlier work using machine learning to analyse MRI scans and predict tumour boundaries, but the leap to live intraoperative guidance required advanced computer vision algorithms trained on thousands of surgical videos.

Health officials have not yet disclosed the specific AI platform used or plans for broader adoption, but the success has prompted discussions about expanding the technology to other complex brain surgeries and potentially to other surgical specialties. The case is expected to be published in a peer-reviewed medical journal in the coming months.

While the procedure is a promising demonstration, experts caution that widespread implementation will require rigorous clinical trials, regulatory approval, and training for surgical teams. Questions also remain about the AI's performance in varied tumour types and anatomical locations beyond this single successful case.

Why This Matters

Patient impact: This technique could reduce the risk of permanent vision loss and other neurological deficits in brain tumour patients, improving quality of life.
Broader significance: It demonstrates that AI can be safely and effectively deployed in real-time during the most delicate surgeries, opening the door to similar applications in other fields (e.g., spinal, cardiac, or ophthalmic surgery).
What happens next: NHS Trusts and international surgical centres will likely accelerate trials of intraoperative AI; regulatory bodies may need to update guidelines for AI-assisted medical devices.

Background

For years, AI has been used in neurosurgery primarily for pre-operative planning — analysing scans to map tumour margins and critical structures. The leap to intraoperative use is much harder because the brain shifts during surgery (brain shift) and tissue appearance changes under the microscope. Previous attempts at AI-assisted surgery have been limited to laboratory settings or simpler, less critical procedures. This case, performed at one of the UK's leading neurology hospitals, represents the first successful clinical application of real-time AI guidance for a brain tumour resection. The technology relies on computer vision models trained on thousands of hours of surgical video to recognise anatomical landmarks.

Key Perspectives

[Surgical team at NHNN]: They view the AI as a co-pilot that enhances the surgeon's precision, especially in identifying structures invisible to the naked eye. They are keen to refine the system and expand its use.
[Patient Rhys Hibbert]: He has expressed gratitude for a procedure that preserved his vision, which a traditional approach might have compromised. He hopes others will benefit from the innovation.
[Critics and sceptics]: Some neurosurgeons and medical device regulators urge caution, noting that a single successful case does not prove safety or efficacy. Concerns include over-reliance on AI, potential for algorithmic errors in atypical anatomy, and the need for fail-safe human oversight. Long-term outcomes are not yet known.

What to Watch

Publication of the surgical technique in a peer-reviewed journal, expected within six months, which will disclose the AI platform and methodology.
The number of additional AI-assisted brain tumour surgeries performed at NHNN and other centres, indicating whether the procedure is reproducible.
Regulatory updates from bodies such as the UK Medicines and Healthcare products Regulatory Agency (MHRA) on approval pathways for intraoperative AI systems.

ZOTPAPER