AI’s Role in Breast Cancer Screening
A recent study has shown that using artificial intelligence in breast cancer screenings may lower the diagnosis rate by 12% in the following years and contribute to detecting cancers earlier. This trial, the largest of its kind, included around 100,000 women in Sweden who underwent mammography screenings. They were split into two groups: one utilizing AI support and the other receiving standard assessments from two radiologists between April 2021 and December 2022.
The AI system was designed to evaluate mammograms, categorizing lower-risk cases for single readings while assigning higher-risk cases to double readings by radiologists. Additionally, it highlighted any suspicious findings to aid the doctors.
The findings, published in The Lancet, indicated that AI-assisted mammography reduced cancer diagnoses to 1.55 per 1,000 women, compared to 1.76 in the control group. Notably, over 81% of cancers in the AI group were detected during the screening, whereas just under 74% were caught in the standard group. Moreover, there were 27% fewer aggressive cancer subtypes in the group using AI.
Dr. Kristina Lång, leading the research at Lund University, mentioned that while AI could be beneficial for early cancer detection, certain precautions need to be taken. She highlighted that integrating AI into breast cancer screening can alleviate some pressures on radiologists while enhancing early-stage detection, especially for more aggressive cancers.
Despite these benefits, Lång emphasized the importance of careful implementation of AI in healthcare. She noted that proper, tested tools should be utilized and ongoing monitoring is essential to ascertain how AI impacts various regional and national screening programs over time.
With breast cancer being a leading cause of death among women aged 35 to 50, and over 2 million new cases diagnosed globally each year, the study’s outcome is significant. However, the researchers caution against completely relying on AI, as human oversight with at least one radiologist reading the results remains crucial.
Dr. Sowmiya Moorthie from Cancer Research UK described the results as promising but underscored the need for caution. While AI can enhance mammogram interpretation efficiency, there are worries about potentially overlooking some cancers. This research, being confined to a single center, necessitates additional studies to confirm its efficacy in life-saving contexts.
Moorthie further remarked, with an anticipated rise in cancer diagnoses, innovations like AI will be critical to enhancing the NHS. Yet, thorough evaluations are essential to ensure that these advancements assist rather than harm patients.
Simon Vincent, chief scientific officer at Breast Cancer Now, indicated that this trial highlights AI’s significant potential to support radiologists in breast cancer screenings. He pointed out that early detection through screening is vital for successful treatment, and the study shows promise for improving diagnostic timelines. Ongoing trials in the UK exploring AI’s integration into the NHS breast screening system will be crucial to understand how to best utilize these technologies for early cancer detection.
