Volumetric breast density characteristics as determined from digital mammograms

Area of Science:

• Radiology and Medical Imaging
• Biomedical Engineering
• Quantitative Imaging

Background:

• Accurate breast volume and density estimation are crucial for mammography analysis.
• Previous methods often lack automation and reproducibility.
• Digital mammography offers potential for quantitative tissue analysis.

Purpose of the Study:

• To present an automated, reproducible algorithm for estimating breast volume and volumetric breast density from digital mammograms.
• To analyze the relationship between volumetric breast density, breast volume, and age in a large cohort.
• To validate the algorithm's performance on a substantial dataset.

Main Methods:

• Development of an automated algorithm using digital mammograms (CC view only).
• Algorithm calibration based on digital image signal versus tissue properties.
• Estimation of compressed breast thickness using an empirical model accounting for compression force.
• Application to 55,087 images from 15,351 women.

Main Results:

• Mean volumetric breast density was 30% and mean breast volume was 687 cm³.
• Strong correlations observed between left and right breast density (r=0.92) and volume (r=0.91).
• Volumetric density decreased with age (45% at 35 years to 25% at 75 years), with a slight increase to 30% at 80 years.
• Annual changes: density decreased by ~2% per year, while volume increased by ~2% per year.

Conclusions:

• The developed algorithm provides a reproducible and automated method for breast volume and density quantification.
• Age is a significant factor influencing volumetric breast density and breast volume.
• The findings support the use of quantitative imaging biomarkers in mammography for understanding breast composition changes.