AI identifies early breast cancer that NHS doctors missed

The NHS has successfully trialled an AI tool that detected breast cancer in 11 women missed by doctors.

During the test, the AI tool ‘Mia’ flagged 55 scans out of 10,000, that doctors concluded were normal, for a second review.

A usual screening will see two doctors read a scan for signs of cancer, and if they disagree, a third doctor will examine it.

Now, Dr Gerald Lip, clinical director of North-east Scotland breast screening programme, says the AI can act as a third eye: “It’s a safety net,” he said on the BBC 4 Today Show.

“We triggered the AI after two humans said it was normal, and then the AI will say: there’s something else there, do you want to have a look.”

11 out of the 55 women called back did in fact have cancer, he said.

“These are real women, real lives, and real cancers that would not have been detected for three years.”

Lip also hailed the use of the tool for reducing staff workload and burnout.

“The incidents of [breast] cancer is six-to-eight out of a thousand, and a regular NHS consultant like myself reads 5000 mammograms a year to find 30 to 35 cancers,” he said. “There’s a lot of normal mammograms in there.”

As a result of the trial, AI managed to reduce staff workload by 30%, Lip noted, adding that he believes the tech could one day free him up to spend more time with patients.

“I see Mia as a friend and an augmentation to my practice.”

While the trial is deemed a success, Mia’s results will still necessitate human review. With no access to patient history, it will flag already identified cysts, and due to current health regulations, it can’t use machine learning to evolve on the job.

Next, Lip hopes to take the trial to a national scale.

At the end of last year, UK hospital Hereford County Hospital completed its first trial of a new AI diagnostic tool aimed to help stroke victims and other heart problems, read about it here.