Lesion Detectability and Masking Disparity Assessment in Breast Tomosynthesis Across Diverse Populations Using...

Area of Science:

• Medical Imaging
• Radiology
• Breast Cancer Screening

Background:

• Breast density is a known factor affecting mammogram accuracy.
• Disparities in breast cancer screening outcomes exist across racial groups.
• The specific role of breast density versus systemic bias in these disparities is unclear.

Purpose of the Study:

• To isolate the effect of breast density on lesion detectability across different racial subgroups.
• To investigate the relationship between race, breast density, and mammographic lesion detection.
• To quantify the contribution of breast density to race-associated differences in detectability.

Main Methods:

• Retrospective case-control study of 902 women using raw digital tomosynthesis projections.
• In-silico insertion of simulated masses and microcalcifications into mammographic images.
• Detection performance assessed using Channelized Hotelling Observers (CHOs).
• Regression and causal mediation analyses to examine race, density, and detectability.

Main Results:

• Lesion detectability significantly decreased with increasing breast density (BI-RADS A to D).
• Area under the receiver operating characteristic curve (ROC AUC) for masses decreased from 0.93 to 0.85, and for microcalcifications from 0.85 to 0.78.
• Breast density accounted for 38-55% of observed race-associated differences in lesion detectability.
• Slightly higher detectability in Non-Hispanic Black women was largely explained by density differences.

Conclusions:

• Breast density is the dominant factor influencing lesion detectability in mammography.
• Findings support the development of personalized screening strategies accounting for individual breast density.
• Calibrated detection models are needed to address density-related challenges in cancer screening.