By Niall Phelan, Medical Physicist, BreastCheck
We recently hosted a workshop to discuss developments in the use of Artificial Intelligence (AI) in breast cancer screening.
The workshop aimed to discuss:
- the potential benefits and opportunities of harnessing the technology, and the associated challenges
- the experiences of using AI in breast screening internationally
- practical aspects of implementation of the technology in BreastCheck
- issues around patient understanding and acceptance of the technology
- legal and ethical concerns.
We welcomed international experts on AI and commercial providers of AI solutions for breast screening to discuss the fundamentals of AI development, practical and technical challenges of implementation and resource requirements. We discussed the outcome of clinical trials, AI platforms operating in other countries and different approaches to models of implementation.
We learned that the use of AI in breast screening can offer many benefits, including:
- Improved accuracy: AI algorithms can analyse mammograms with high accuracy, potentially reducing false positives and false negatives. These algorithms can detect patterns and abnormalities in images that may be missed by human radiologists.
- Speed and efficiency: AI can process mammograms much faster than humans, leading to quicker results.
- Assistance to radiologists: AI can act as a second opinion for radiologists, helping them in their decision-making process.
- Personalised risk assessment: AI algorithms can analyse various risk factors, including breast density and medical history to provide personalised risk assessments for breast cancer. This information can help personalise screening and prevention strategies to individual patients.
- Data analysis and research: AI can analyse large datasets of mammograms and patient records to identify trends, risk factors, and treatment outcomes, leading to advancements in breast cancer detection and treatment.
- Workforce issues: AI-powered screening tools can potentially improve access to breast cancer screening and adherence to screening rounds in the context of a shortage of radiologists, and allow radiologists to concentrate time on assessment and care of women who have breast cancer detected.
We also learned that it will be essential to ensure that AI technologies are rigorously tested and validated before widespread adoption to ensure patient safety and prove effectiveness. It will be important to forge patient trust in the technology and the enhanced role of AI in the detection and diagnosis of breast cancer. Ethical considerations, such as informed consent, patient privacy, algorithm bias, explainability and ensuring equitable access will need to be addressed in the development and deployment of AI in breast cancer screening.
There is a definite recognition that AI technology for breast cancer screening has emerged from the research and evaluation stage to clinical implementation and offers potential to the delivery of screening services, and importantly to improved confidence and care to our screening participants.