Researchers have demonstrated the power of artificial intelligence (AI) in breast cancer detection.
In a ground-breaking study published in The Lancet Oncology, a Swedish team found that AI-supported mammograms detected 20% more breast cancers compared to the routine double readings conducted by two radiologists.
One of the most promising aspects of this research is that the AI system didn’t cause an increase in false positives.
Traditional mammography screenings have long relied on the expertise of radiologists to identify potential cancers, and it can take more than 10 years to train a radiologist capable of interpreting mammograms.
This process almost halves the screen-reading workload – a key consideration in the current climate. There is a shortage of breast radiologists in many countries, including a shortfall of around 41 (8%) in the UK and about 50 in Sweden.
However, this study, which represents a significant advancement, introduces AI as a supportive tool to enhance accuracy and patient safety of screenings.
Despite the numerous retrospective studies showing the potential of AI in mammography, this is the first randomised trial to validate these results.
The study involved women aged 40 to 80 years eligible for mammography screening at four sites across Sweden.
Participants were randomly assigned to receive AI-supported screenings or standard double readings by radiologists without AI assistance.
The AI system, known as Transparent version 1.7.0, provided a malignancy risk score for each examination.
Based on this score, the system then guided radiologists to decide whether further screening was necessary.
Throughout the study, 80,033 women participated, with roughly half assigned to each group.
The AI-supported screening group detected 244 cancers out of 39,996 participants, while the standard screening group identified 203 cancers out of 40,024 participants. Although the ratio was insignificant, the AI-supported group had a slightly higher cancer detection rate.
Crucially, introducing AI substantially reduced the screen-reading workload by an impressive 44.3%.
This finding suggests that AI can significantly streamline the review of mammography images, allowing radiologists to focus more on complex cases.
The authors emphasised that AI-enhanced mammography screening is a safe and viable approach. Notably, the trial continues to assess the primary endpoint, the interval cancer rate, over a two-year follow-up period.
Lead author Dr Kristina Lång from Lund University, Sweden, said: ‘We still need to understand the implications on patients’ outcomes, especially whether combining radiologists’ expertise with AI can help detect interval cancers that traditional screening often misses, as well as the cost-effectiveness of the technology.’
The final trial results – and whether AI’s use in mammography screening is justified – are not expected for several years.
The Swedish Cancer Society, the Confederation of Regional Cancer Centres and the Swedish governmental funding for clinical research (ALF) financially supported the research.
Meanwhile, a recent study published in the Annals of Internal Medicine questioned the value of breast cancer screening in older women. It found overdiagnosis of up to 31% of breast cancer cases among women aged 70 to 74 who underwent mammograms.
In an editorial accompanying the study, oncologists Otis Brawley and Rohan Ramalingam observe that physicians need new tools to identify the stage and severity of cancer objectively.
