AI shows 'promising potential' in breast cancer-screening trial

Human radiologists’ detection rate matched by screenings using artificial intelligence in Swedish study

AI-assisted screenings could reduce workload for radiologists and speed up diagnosis, researchers found. EPA

Artificial intelligence (AI) has shown a capacity to “safely” evaluate breast cancer screening images, displaying a similar rate of detection as human radiologists, a study has shown.

Published in the journal Lancet Oncology, the research placed the AI technology in direct comparison with standard human assessment, marking a significant departure from past studies which merely reviewed scans previously made by medical professionals.

More than 80,000 women from Sweden, averaging 54 years of age, took part in the trial.

Half of their mammograms – X-ray images of the breast – were analysed by two radiologists, deemed as standard care, while the other half were first assessed by the AI tool, followed by one or two radiologists.

Cancer was detected in 244 patients via AI-assisted screenings, compared to 203 standard diagnoses.

The AI did not result in an increased number of “false positives” – instances where a scan is incorrectly identified as abnormal – maintaining a 1.5 per cent rate consistent with that of the human assessments.

Lead author Dr Kristina Lang, from Lund University in Sweden, addressed the implications of the study, saying: “The greatest potential of AI right now is that it could allow radiologists to be less burdened by the excessive amount of reading … easing the pressure on workloads and enabling radiologists to focus on more advanced diagnostics while shortening waiting times for patients.”

The study has highlighted the potential for reduced radiologist workloads and shorter waiting times for patients. PA

The study revealed that the integration of AI could cut the screening workload nearly in half, with 36,886 fewer screen readings required by radiologists in the AI-assisted group, translating to a 44 per cent reduction.

However, Dr Lang cautioned against premature introduction of AI in mammography screenings.

“These promising interim safety results should be used to inform new trials and programme-based evaluations to address the pronounced radiologist shortage in many countries, but they are not enough on their own to confirm that AI is ready to be implemented in mammography screening,” Dr Lang said.

NHS officials have already begun to explore the potential role of AI in breast screening.

An NHS spokesman said: “This research is very encouraging and plans are under way to assess the best ways of implementing this technology into the NHS breast screening programme.”

Dr Katharine Halliday, president of the Royal College of Radiologists, said AI could “save clinicians time by maximising our efficiency, supporting our decision-making and helping identify and prioritise the most urgent cases”.

The continuing study continues to evaluate the long-term potential of AI tools in detecting cancers between screenings, with the final results expected in a few years' time.

“We look forward to the final results of this exciting Swedish trial to understand if AI can help improve breast cancer screening and increase capacity in the future,” said Dr Kotryna Temcinaite, head of research communications at Breast Cancer Now.