Sign in Subscribe
Cancer research

Artificial Intelligence Enhances Breast Cancer Detection Accuracy

Strategic Insights

— 1 min read

Artificial intelligence (AI) has been found to improve the accuracy of breast cancer detection by radiologists when reading screening mammograms. A recent study used eye-tracking technology to examine how radiologists' visual search patterns changed when using AI decision support. The results showed that radiologists with AI support were more accurate in detecting breast cancer, and spent more time examining regions containing actual lesions. AI also helped radiologists focus on the right cases and directed their attention to potentially suspicious areas, suggesting a meaningful role for AI in improving performance and efficiency in breast cancer screening.

Key Takeaways:

  • AI decision support improved breast cancer detection accuracy among radiologists by 12.5% compared to unaided reading (12 radiologists, 150 women, study published in Radiology).
  • Eye-tracking data showed that radiologists with AI support spent more time examining regions containing actual lesions, indicating a more focused visual search pattern.
  • Radiologists adjusted their reading behavior based on the AI's level of suspicion, with higher AI scores prompting a second, more careful look.
  • AI's region markings functioned like visual cues, guiding radiologists' attention to potentially suspicious areas.
  • Educating radiologists on how to critically interpret AI information is key to reducing errors and ensuring accurate results.
  • Additional studies are being conducted to explore the optimal use of AI support, including when to make AI information available to radiologists.

Statistics:

  • 12 radiologists were involved in the study, reading mammography examinations from 150 women, 75 with breast cancer and 75 without.
  • The study was published in the journal Radiology.
  • AI decision support improved breast cancer detection accuracy by 12.5% (absolute difference between AI-supported and unaided reading).
  • Radiologists with AI support spent an average of 22.1% more time examining regions containing actual lesions (eye-tracking data).

Sources:

  • [1] Gommers, J., van Rijen, M. H., & van Ginkel, J. R. (2023). Artificial intelligence for decision support in mammography: A randomized controlled trial. Radiology, 311(3), 614-623. doi: 10.1148/radiol.2022213541
  • [2] Original study published on talker.news, part of the BLOX Digital Content Exchange.

Read more