New study finds AI tool outperforms experts in spotting breast cancer!

AI analysis flagged up to 13 per cent more cases than doctors had identified. – Rui Vieira/PA

Dear Commons Community,

A new study has found that artificial intelligence is significantly better at spotting breast cancer than humans.

The research, published in Nature, found that an AI device which utilized sophisticated image recognition technology to scan mammograms detected up to 13 per cent more breast cancers than radiologists.

Some 20 per cent of tumors are missed during breast screening, but the study, conducted by Imperial College London and Kheiron Medical Technologies, raises hope that this rate could one day be greatly reduced.  As reported by The Telegraph.

The researchers used the AI device, known as Mia, as an extra ‘reader’ to analyse the mammograms of more than 25,000 women over three phases – two pilot phases and one live roll-out – across four screening sites in Hungary.

The mammograms were first examined as usual by two radiologists, before being studied by Mia, which flagged potential false negatives by identifying subtler signs of cancerous tissue that had been deemed healthy by the doctors.

The results showed that the AI delivered improvements in cancer detection rates of 5 per cent, 10 per cent and 13 per cent, respectively, compared with the average reading analysed by at least two radiologists.

It also found that 83 per cent of the additional cancers detected by the device were invasive, meaning that they have the potential to spread to other parts of the body – further highlighting AI’s potential to tackle cancer by providing crucial early detections.

“Our study shows that using AI can act as an effective safety net – a tool to prevent subtler signs of cancer falling through the cracks,” said study co-author Dr Ben Glocker, from Imperial’s Department of Computing.

“Seeing first hand that the use of AI could substantially reduce the rate of missed cancers in breast screening is massive, and a major boost for our mission to transform cancer care with AI technology.”

One in six people will die of cancer, figures show. In 2020, 2.3 million women across the globe were diagnosed with breast cancer, and 685,000 women were killed by the disease, according to the World Health Organization.

“The key question now is how we can justify not using Mia in breast screening when there is such a dramatic improvement in cancer detection,” said Dr Peter Kecskemethy, CEO of Kheiron.

Around the world, healthcare systems are under immense strain, with many cracks exacerbated by the Covid-19 pandemic, such as increased delays in screening and diagnosis.

According to GE Healthcare, for each day the United States spent in lockdown around 100,000 mammograms were added to the nation’s screening backlog.

In 2020, 2.3 million women across the globe were diagnosed with breast cancer, and 685,000 women were killed by the disease, according to the World Health Organization. – ilbusca/GETTY IMAGES

Radiology departments are also struggling, with just 2 per cent of UK radiology departments having the capacity to fulfil their imaging reporting requirements within contracted hours, according to the Royal College of Radiologists.

Three out of four imaging leaders believe that the National Health Service (NHS) does not employ enough staff to ensure “safe and effective care,” according to The Royal College of Radiologists.

Kheiron emphasises that doctors are under immense pressure, especially in screening units, where there is a large volume of cases and a high expectation to go through backlogs. Whereas doctors’ levels of performance are variable, AI has consistent performance, the company said.

It added that it is concerned by the high levels of inaccurate imaging in radiography. Figures show that inaccurate breast positioning is responsible for 47 per cent of technical failures, further increasing the backlog of mammograms and misdiagnosis of breast cancer.

This is where harnessing the power of AI technology, which can analyse images at more intricate pixel depths that doctors can’t, could be crucial, experts say.

“We see the future as humans and AI working together,” said Tobias Rijken, co-founder of Kheiron Medical.

“The way that I always like to think about it is that humans and AI have their own different strengths and I think our paper showed that,” he added.

“I think it’s the two working in tandem that gives you the best performance, if you know how to optimally combine the two workflows.”

Blending human and AI expertise is fast becoming the norm in many human endeavors!

Tony

Comments are closed.