Artificial Intelligence for Digital Breast Tomosynthesis Screening with and without Prior Examinations in BreastScreen Norway
Radiology: Artificial Intelligence, 2026
Abstract
Introduction: Artificial intelligence (AI) systems are increasingly being evaluated for their potential in breast cancer screening, but their performance with digital breast tomosynthesis (DBT) and the added value of prior examinations remains an area of active investigation.
Methods: This study evaluated an AI model for detecting breast cancer on digital breast tomosynthesis screening examinations, with and without the inclusion of prior examinations, using data from BreastScreen Norway.
Results: The AI model demonstrated performance similar to that of radiologist double reading for breast cancer detection with digital breast tomosynthesis. The study found minimal additional benefit from incorporating prior examinations into the AI analysis.
Conclusion: AI shows promise as a tool comparable to double reading by radiologists in DBT screening, and the routine inclusion of prior examinations may offer limited added value for AI-based detection.
DOI: Read full study
