E2044. Implications of Change in an AI-Derived Case Score on Outcomes of Sequential Years of Digital Breast Tomosynthesis Screening
  1. Samantha Zuckerman; University of Pennsylvania Health System
  2. Christine Edmonds; University of Pennsylvania Health System
  3. Elizabeth McDonald; University of Pennsylvania Health System
  4. Jennifer Tobey; University of Pennsylvania Health System
  5. Julia Birnbaum; University of Pennsylvania Health System
  6. Susan Weinstein; University of Pennsylvania Health System
  7. Emily Conant; University of Pennsylvania Health System
Screening with digital breast tomosynthesis (DBT) has been shown to improve both sensitivity and specificity compared to screening with digital mammographic alone. A DBT-derived, artificial intelligence (AI) algorithm (iCAD ProFound AI V2.0, iCAD Nashua, NH. has shown promising results in improving both screening accuracy and decreasing reading times. The purpose of this study is to demonstrate examples of both case and lesion-level findings after sequential rounds of DBT screening with the incorporation of the DBT-derived AI algorithm. Examples of changes over sequential rounds of screening of both AI-CAD lesion-level and case scores will be demonstrated and correlated with the key clinical outcomes of screening (TP, TN, FP and FN cases and rates).

Educational Goals / Teaching Points
Incorporation of a DBT-derived, AI algorithm may alert radiologists to both the case-level likelihood of malignancy as well as the location of specific, suspicious lesions. In addition, an increase in case-level and/or lesion-level score over sequential screens may aid radiologists in identification of screen-detected cancers and identify patients at higher risk for developing breast cancer.

Key Anatomic/Physiologic Issues and Imaging Findings/Techniques
Examples of DBT AI-CAD detections and missed opportunities will be demonstrated by breast density, lesion type (calcification, mass, distortion) and histology when available, and implications of change case scores over time.

A DBT-derived, AI algorithm may help radiologists identify potentially suspicious lesions. Increased case scores over time may also alert radiologists to lesions that may have been missed and may also identify patients who are at increased risk of developing breast cancer.