AI Boosts Breast Cancer Detection by 10% in UK Screening Program
Artificial intelligence (AI) is significantly improving breast cancer detection rates and reducing workloads in the United Kingdom’s National Health Service (NHS) screening program, according to findings from the GEMINI study published in Nature Cancer. The integration of AI into screening workflows led to a 10.4% increase in cancer detection and workload reductions of up to 31%.
Background: The Challenge of Breast Cancer Screening
In the UK, over 2 million mammograms are performed annually, inviting women aged 50 to 70 for screening every three years. Currently, two radiologists independently review each mammogram to minimize missed cancers. However, detecting subtle signs of breast cancer can be challenging, leading to false negatives – where cancers are missed – and false positives, resulting in unnecessary recalls for further investigation. Approximately 20% of cancers are missed with the current double-reading process. For every five women recalled for further assessment, only one is ultimately diagnosed with cancer, causing anxiety and potentially invasive tests for those with benign findings.
The GEMINI Study: AI Integration and Evaluation
The GEMINI study prospectively evaluated the impact of integrating Live AI with Mammography Intelligent Assessment (Mia) v.3 into breast cancer screening across a region of the United Kingdom. Researchers assessed 10,889 women, with scans evaluated by both the AI tool and a human reader. Cases where the AI and human assessments differed underwent further review by a human expert.
Key Findings: Improved Detection and Reduced Workload
The study revealed that disagreements between the AI tool and routine double reading led to the detection of 11 additional cancers, resulting in a 10.4% improvement in cancer detection – equivalent to one additional cancer detected per 1,000 patients screened. The recall rate was reduced by 0.8%, and workloads were reduced by up to 31%. Fewer patients were recalled for biopsies due to false-positive results, leading to reduced healthcare costs, resource utilization, and patient stress.
Variations in AI implementation further demonstrated potential workload savings of up to 36%, alongside improvements in cancer detection rate, recall rate, positive predictive value, sensitivity, and specificity. The time to notify women with detected cancer was also significantly reduced, from 14 days to 3 days, a change the study authors considered “hugely significant” as earlier detection increases the likelihood of successful treatment.
Future Directions and Ongoing Research
Despite the promising results, the UK National Screening Committee has not yet recommended the widespread use of AI in the NHS breast screening program, citing previous concerns about the quality and quantity of evidence. However, the GEMINI study provides high-quality evidence supporting the integration of AI into screening workflows. Researchers are now expanding this work in the upcoming EDITH trial, which will evaluate AI use in breast cancer screening across the entire United Kingdom.
Clarisse Florence de Vries, PhD, MSc, lead author of the study from the Glasgow Lab for Data Science & AI, Public Health, School of Health and Wellbeing, University of Glasgow, stated that the findings demonstrate that AI use can be tailored to local healthcare needs to enhance service delivery.
Disclaimer: This article is for informational purposes only and does not constitute medical advice.