How AI could help make brain surgery safer

STORY: Could artificial intelligence be used to make brain surgery safer?

At this university in London, trainee surgeon Danyal Khan is taking part in a mock operation during which he's assisted by a real-time video feed, as is typical in brain surgery.

But what's new here is that the footage is being analyzed by AI to help Khan better understand what he's seeing.

The AI system, which is under development at University College London (UCL), highlights sensitive or critical structures in the brain.

Neurosurgeon Hani Marcus believes it has the potential to make brain surgery safer and more effective:

"So I'm very bullish that in the medium to long term, the A.I. will be helping lots of surgeons do lots of operations better than they otherwise can."

Marcus says the AI system analyzed video of more than 200 pituitary gland tumor operations, and gained around 10 years-worth of experience in a fraction of the time.

That knowledge means the AI can now not only help navigate to the correct area of the brain, but also know what should be happening at any stage of the procedure, making it a valuable training aid.

"So, what we're really trying to do is apply AI or artificial intelligence to support surgeons doing brain tumor surgery at the base of the brain. And what this practically entails is us training the AI with hundreds of videos, telling it, if you like, what structures are what and then at some point over that period, the AI becomes really good itself at recognising things, and able to support other surgeons who're perhaps less experienced in advising them what to do next."

Assistant Professor of Robotics and A.I. Sophia Bano explains how that might look in a real operation:

"There can be scenarios where clinicians, unintentionally, are very close to a very critical structure such as the optic nerve. This can have, any damage or a slight more pressure on the optic nerve, can have long term complications on the patient side. So, this whole tool will alert the surgeon during the procedure if there is any risk of potential complication so they can recalibrate themselves during the procedure."

Khan, who was also involved in developing the software that is now helping him learn, says the A.I. system has been a valuable tool in his training.

It could also provide him with step-by-step guidance during a procedure, similar to having a senior surgeon standing over your shoulder.

"I think as I progress as a surgeon, there might be stages where I wonder, you know, have I done enough of a particular part of the procedure and should I move on? And actually having that sort of assistant in the background as a reassurance to look at and say, 'well, yeah, actually, at this stage, out of the hundreds of videos of experts that this algorithm has watched, the experts would probably start moving on to the next phase'. It's a useful double check."

According to UCL, the system could be ready to be used in operating theaters within two years.