Surgeons at University College London Hospital (UCLH) in the United Kingdom have successfully removed a brain tumour with direct help from artificial intelligence (AI) during the operation. They describe it as the world's first brain-tumour removal carried out with AI assistance. According to the medical team, the 48-year-old patient, named Rhys Hilbert, had a tumour located in a part of the brain where even a minor error risked damaging crucial blood vessels and nerves.

The doctors noted that had the operation not been performed, Hilbert could have lost his eyesight. During the procedure, the AI system analysed live video to identify the positions of important blood vessels and nerves in the brain. It also flagged comparatively safer areas for removing the tumour, guiding the surgeons. Hilbert's tumour sat very close to the carotid artery, which supplies blood to the brain, and the optic nerve, which controls vision. Over time, the tumour had begun pressing on the optic nerve, reducing his peripheral vision. He started bumping into objects and also suffered from severe fatigue and dizziness.

After the operation, Hilbert's vision improved considerably. He said that when he opened his eyes after surgery, he felt he could see everything around him, and that he had not seen so clearly in a year. Along with the gradual recovery of his sight, the side effects he had previously experienced also eased.

Initially, Hilbert did not need surgery. Later, when his condition deteriorated, the doctors decided to operate. When he was offered the chance to take part in this new type of AI-assisted surgery, he agreed. During the operation, an endoscope was inserted through the nose to reach the base of the skull. A narrow camera at the tip of the endoscope provided a view of the surgical site. The tumour lay hidden behind bone and a tough lining layer, making it difficult to find the safest route for removal, and such procedures carry a risk of complications. According to the doctors, in 25 to 50 percent of cases it may not be possible to remove the entire tumour.

In this situation, the AI system helped the surgeons. It analysed the live video feed from the endoscope camera during the operation and also tracked the movement of the surgical instruments. It identified where the brain's important blood vessels and nerves were most likely to be, displaying them on screen, and pinpointed the comparatively safer areas for removing the tumour. The technology works much like the face-recognition feature of a smartphone, except that instead of faces, it detects the brain's vital and often hidden structures.

Researchers at University College London trained the AI system using videos of hundreds of pituitary tumour operations, in which blood vessels, nerves and other important parts were individually marked. The AI was taught to recognise key physical structures by analysing that data. Professor Hani Marcus, a surgeon involved in the procedures, said the AI system had been trained on so much surgical data that most surgeons would not perform that many operations in their entire careers. It can therefore act as a specialist's "second pair of eyes" during an operation. A key feature of the system is that it can analyse live information even as surgery is underway. The AI did not make any decisions; the surgeons decided when and where to operate, while the AI simply supported them with the necessary information.

Source: BBC