Automated volumetric breast density estimation: a comparison with visual assessment

Area of Science:

• Radiology
• Medical Imaging
• Breast Health

Background:

• Breast density assessment is crucial for mammography interpretation and breast cancer risk stratification.
• Visual assessment of breast density is subjective and can vary between radiologists.
• Automated methods offer objective and reproducible breast density measurements.

Purpose of the Study:

• To compare automated volumetric breast density (VBD) measurements with radiologist visual assessment using BI-RADS categories.
• To identify factors influencing the agreement between automated VBD and visual assessment.

Main Methods:

• Retrospective analysis of 193 screening mammograms with negative findings.
• Qualitative BI-RADS density categorization by three radiologists.
• Automated measurement of VBD (% breast density) and volumetric density grade (VDG).
• Comparison of VDG with visual assessment and analysis of influencing physical factors.

Main Results:

• Good agreement was found between radiologist visual assessment and VDG (ICC = 0.757).
• VBD demonstrated a strong positive correlation with visual assessment (Spearman's ρ = 0.754, p < 0.001).
• Significant differences in VBD and x-ray tube target were observed between agreement and disagreement groups (p = 0.02 and p = 0.04).

Conclusions:

• Automated VBD measurement is a reliable and objective method for quantifying breast density.
• Physical factors, such as the x-ray tube target, can impact the agreement between automated VDG and visual density assessment.