Validation Using Breast MRI

Breast density has been shown to reduce the sensitivity of mammography and increase a woman’s risk of developing breast cancer. Accurate methods for measuring breast density are, therefore, essential to optimize the breast cancer screening process and enhance objective clinical decisions regarding supplemental imaging. Radiologists typically judge breast density visually by assessing the area of dense tissue from a 2D mammogram, an imprecise and subjective method known to suffer from both inter- and intra-reader variability.

Fully automated software, such as VolparaDensity, has now been developed to estimate the volume of fibroglandular (dense) tissue in the breast from full-field digital mammograms (FFDM). To demonstrate that such volumetric estimates are accurate, several researchers have published validation studies comparing volumetric breast density estimated from FFDM, to volumetric measurements estimated from breast MRI.

Independent Research Shows High Correlation to Breast MRI

Researchers compared VolparaDensity to MRI, in the largest validation study to date [1]. Mammograms and MRI images were obtained from women undergoing high-risk breast cancer screening at the Radboud University Nijmegen Medical Centre. The MRI and mammography exams were done within two months of each other, to minimize potential changes in breast density over time. An automated segmentation method for estimating breast density worked effectively on 74.4% of the MRI studies. From the corresponding 186 digital mammographic studies (i.e. 680 images that included both CC and MLO views), volumetric breast density estimates were obtained using VolparaDensity software and validated against the MRI results. VolparaDensity estimates were highly correlated with the MRI measurements (r= 0.93, 0.97 and 0.85, for volumetric breast density, breast volume, and fibroglandular volume, respectively).

This study supports the findings of two other validation studies, including another recent publication by the University of California, San Francisco (UCSF) that compared several volumetric breast density methods (VolparaDensity, Single-energy X-ray Absorptiometry (SXA) and Quantra) with breast MRI [2]. The study consisted of MRI images (left breast) taken within a year of the corresponding mammographic exam (left CC view only included in the study), from 99 women undergoing screening mammography at the UCSF medical center. VolparaDensity showed good correlations with MRI for volumetric breast density, breast volume and fibroglandular volume (r = 0.85, 0.95, and 0.79, respectively). Although a much smaller dataset was used (26 women), the third validation study also found a high correlation between VolparaDensity and breast MRI for fibroglandular volume (0.94). All three studies showed high correlation to Breast MR, highlighting the accuracy of VolparaDensity in measuring volumetric breast density.

Same Woman with Dense Breasts Examined by MRI and Mammography


In the video above, a large proportion of fatty tissue is evident for this woman assessed with dense breasts (BI-RADS 4). A visual assessment of the mammography study below for the same woman, gives little indication of the volume of fatty tissue present while VolparaDensity assessed the volumetric density as 22.2% (average of the left and right views) which corresponds to a BI-RADS 4 using the Volpara Density Grade (VDG) volumetric thresholds.



The MRI exam shows there is a lot of fatty tissue in dense breasts which isn't apparent in mammograms.

 [1] Gubern-Mérida, A., Kallenberg, M., Platel, B., Mann, R.M., Martí, R. and Karssemeijer, N. (2014) Volumetric Breast Density Estimation from Full-Field Digital Mammograms: A Validation Study. PLoS ONE; 9: e85952

[2] Wang, J., Azziz, A., Fan, B., Malkov, S., Klifa, C., Newitt, D., Yitta, S., Hylton, N., Kerlikowske, K. and Shepherd, J.A. (2013) Agreement of Mammographic Measures of Volumetric Breast Density to MRI. PLoS ONE; 8: e81653

[3] Highnam, R., Brady, M., Yaffe, M., Karssemeijer, N., and Harvey, J. (2010) Robust Breast Composition Measurement – Volpara™. IWDM 2010, LNCS; 6136: 342-349