AI and the Breast Health Journey, Part 5

Part 5: Can AI Improve Accuracy in Breast Screening?

In our Breast Health Series, we’ve examined various hurdles slowing down progress in breast screening. What is the best approach to overcoming breast density and limits of mammography while ensuring equitable health standards? AI appears to hold the key to new possibilities in breast imaging and care, but the way we implement AI is just as important as the technological capabilities. Part 5 of the Breast Health Series will serve as an overview of the current landscape of breast health and explore the path forward with AI.

Breast density is perhaps the most challenging factor when it comes to breast imaging. Dense breasts can obscure findings on traditional mammograms, so improving alternative secondary screening methods can greatly reduce uncertainty in readings. Ultrasound is a widely available and affordable form of screening, made more viable based on a new AI tool in development. The researchers noted one of the primary functions of healthcare AI is to reduce medical errors, such as misdiagnosis. This works to improve the patient experience and minimize the burden and resource waste across healthcare.

Secondary readings on mammograms are another way to increase accuracy. Double reading of all mammograms would be ideal, but in reality, having 2 radiologists perform readings on every mammogram is very time consuming, and radiologists are already stretched thin. AI as an automated second reader has shown to significantly ease the workload of radiologists while also providing more accurate readings.

AI treated like a teammate for radiologists is the most impactful way to integrate artificial intelligence in breast screening. A recent study shows that combining AI readings with radiologist readings greatly improves the detection of malignant breast architectural distortion, the most difficult type of tumor to detect and the most commonly missed abnormality. In this case, the combination of methods results in improved readings, as AI alone didn’t yield better results than radiologists alone.

In a previous interview, Dr. Amy Patel explains that her practice sees the benefits of AI as a second opinion consult. Overall, regardless of the radiologist’s individual experience level or effectiveness, AI supports the reading process and allows them to work more confidently.

That concludes our series on AI and the Breast Health Journey. What are your thoughts on the use of AI in breast screening? Should it be a standard of care? We’d love to hear your feedback and learn from your experience, please drop us a note in the comment section below.