Breast cancer screening has taken a significant leap forward with the introduction of AI-supported radiologists. A recent study conducted by Kristina Lång at Lund University in Sweden revealed that individuals screened with the assistance of AI were less likely to develop aggressive cancers before their next screening round, offering hope for improved outcomes and potentially saving lives.
The AI-supported approach involves using software that has been trained on a vast database of over 200,000 mammography scans from 10 different countries. This software ranks the likelihood of cancer being present in mammograms on a scale of 1 to 10 based on visual patterns in the scans. Mammograms receiving a score of 1 to 9 are then assessed by one experienced radiologist, while those receiving a score of 10 – indicating a high likelihood of cancer – are evaluated by two experienced radiologists.
Previous studies have already demonstrated the effectiveness of this AI approach, detecting 29% more cancers than standard screening methods without increasing false detection rates. Fiona Gilbert from the University of Cambridge, who was not involved in the trial, described the results as “terrific.”
The recent analysis of over 100,000 women in Sweden, randomly assigned to either standard screening or AI-assisted screening, showed promising results. Women who received AI-assisted screening were 12% less likely to develop interval cancers – aggressive tumors that rapidly develop between screenings.
The success of the AI approach may be attributed to its ability to detect cancers at very early stages that could be overlooked by radiologists. However, further trials are needed to confirm if AI can outperform standard screening methods.
While the study did not focus on specific ethnic groups, ongoing trials, including one in the UK, aim to address this issue. Additionally, research is needed to determine if less experienced radiologists can benefit from using AI.
Kristina Lång anticipates the implementation of the AI approach in south-west Sweden within a few months. However, it may take up to five years for other countries to complete similar trials before widespread adoption. Factors such as cost-effectiveness and the need for radiologist training must also be considered.
It is essential to emphasize that AI should complement, not replace, the role of radiologists in breast cancer screening. Women participating in screening prefer human involvement alongside AI tools, emphasizing the importance of radiologists in the process.
Overall, the integration of AI in breast cancer screening shows great promise in improving outcomes and reducing interval cancer rates. As technology continues to advance, further research and trials will be crucial in maximizing the benefits of AI-assisted screening.
