A FIRST OF its kind randomised controlled trial involving over 80,000 Swedish women has found that artificial intelligence (AI)-supported mammography screenings are as good as two breast radiologists working together to detect breast cancer.

The AI-supported screening detected 20% more cancers compared with the routine double reading of mammograms by two radiologists.

An interim safety analysis of the trial also found that AI-supported screenings can increase efficiency without increasing false positives and almost halves the screen-reading workload.

The findings of the trial have been published today in The Lancet Oncology journal.

Breast cancer screening with mammography has been shown to improve prognosis and reduce mortality by detecting breast cancer at an earlier, more treatable stage.

European guidelines recommend a double reading of screening mammograms by two radiologists to ensure a correct identification.

But there is a shortage of breast radiologists in many countries and it takes over a decade to train a radiologist capable of interpreting mammograms.

In Ireland, this year’s Medical Workforce report from the HSE found that radiology has the lowest level of trainees, as well as the lowest ratio of Non-Consultant Hospital Doctors (NCHDs) to every consultant.

AI has been proposed as an automated second reader for mammograms to reduce workload and improve screening accuracy, but robust evidence had been lacking.

80,033 women aged 40-80 took part in the trial between April 2021 and July 2022, and had undergone mammogram screening at four sites in southwest Sweden.

They were randomly assigned to either AI-supported analysis, where an AI-supported mammogram reading system analysed the mammograms before they were also read by one or two radiologists, or standard analysis performed by two radiologists without AI.

The interim analysis compared early screening performance and screen-reading workload.

AI-supported screening detected 20% more cancers compared with the routine double reading of mammograms by two breast radiologists, with 41 more cancers detected with the support of AI.

The false-positive rate for both arms of the trial was 1.5%.

Overall, AI-supported screening resulted in a cancer detection rate of six per 1,000 screened women, compared to five per 1,000 for standard double reading without AI.

This is equivalent to detecting one additional cancer for every 1,000 women screened.

There were also 36,886 fewer screen readings by radiologists in the AI-supported group than in the control group, resulting in a 44% reduction in the screen-reading workload of radiologists.

Lead author of the report, Dr Kristina Lång from Lund University in Sweden, said the results were “promising” and could inform ways to “address the pronounced radiologist shortage in many countries”.

She added: “The greatest potential of AI right now is that it could allow radiologists to be less burdened by the excessive amount of reading.

“While our AI-supported screening system requires at least one radiologist in charge of detection, it could potentially do away with the need for double reading of the majority of mammograms, easing the pressure on workloads and enabling radiologists to focus on more advanced diagnostics while shortening waiting times for patients.”

However, she warned that the results on their own are “not enough to confirm that AI is ready to be implemented in mammography screening”.

“We still need to understand the implications on patients’ outcomes, especially whether combining radiologists’ expertise with AI can help detect interval cancers that are often missed by traditional screening,” said Lång.

Interval cancers are cancers that are diagnosed in between routine screening episodes.

Estimates suggest that 20-30% of interval cancers that should have been spotted at the preceding screening mammogram are missed, and suspicious findings often turn out to be benign.

Trial results looking at whether the use of AI in interpreting mammography images translates into a reduction in interval cancers are not expected for several years.