Related Experiment Video
Updated: Apr 19, 2026
Tracking the Mammary Architectural Features and Detecting Breast Cancer with Magnetic Resonance Diffusion Tensor Imaging
Published on: December 15, 2014
Mammographic breast density: comparison of methods for quantitative evaluation
Oliver W E Morrish1, Lorraine Tucker, Richard Black
1From the East Anglian Regional Radiation Protection Service (O.W.E.M.), Department of Medical Physics and Clinical Engineering (R.B.), and Cambridge Breast Unit (P.W.), Cambridge University Hospitals NHS Foundation Trust, Cambridge, England; Department of Radiology, University of Cambridge School of Clinical Medicine, Cambridge Biomedical Campus, Cambridge CB2 0QQ, England (L.T., F.J.G.); and Wolfson Institute of Preventive Medicine, Queen Mary University of London, London, England (S.W.D.).
Automated mammographic breast density tools show good agreement with each other, but their results do not strongly correlate with human reader scores. These differences in breast density measurements should be considered for personalized patient imaging.
Area of Science:
- Radiology and Medical Imaging
- Quantitative Analysis
- Biomedical Engineering
Background:
- Mammographic breast density is a key indicator for breast cancer risk.
- Accurate breast density measurement is crucial for personalized screening and risk assessment.
- Automated software tools offer potential for objective and reproducible breast density assessment.
Purpose of the Study:
- To evaluate and compare the performance of two automated software tools (Quantra and Volpara) for mammographic breast density measurement.
- To compare the results from these automated tools with traditional observer-based scores.
- To assess the agreement and correlation between different breast density measurement methods in a large cohort.
Main Methods:
- A dataset of 36,281 mammograms from 8,867 women was analyzed.
- Breast density was assessed using two automated tools (Quantra, Volpara) and by 26 human readers using a visual analog scale.
- Agreement was assessed using Bland-Altman analysis, and correlation with observer scores was calculated using Pearson correlation coefficients.
Main Results:
- High correlation was found between Quantra and Volpara for total breast volume (r=0.97) and percentage density (r=0.78).
- Correlation was lower for fibroglandular volume (r=0.86).
- Automated tool correlations with observer scores were moderate (r=0.55 to 0.63), indicating significant differences.
Conclusions:
- Automated breast density measurement software demonstrates good inter-tool correlation.
- However, automated methods show poor correlation with observer-based scores.
- Discrepancies between automated and observer-based breast density measurements necessitate careful consideration in clinical practice and personalized imaging strategies.
