Feature | April 22, 2015

ECR Research Demonstrates Utility of Volumetric Breast Imaging Data

Series of studies utilize Volpara software to assess density, predict breast cancer risk

April 22, 2015 — Ten abstracts on the uses of volumetric breast imaging data were presented at the 2015 European Congress of Radiology (ECR) meeting, focusing on reducing subjectivity in density and assessment and improving breast cancer risk prediction models. Eight posters and two session presentations highlighted the use of Volpara Solutions’ automated breast density and quantitative breast imaging software tools to provide insight into the impact of quantitative breast density on mammographic sensitivity, utilize volumetric density to improve breast cancer risk prediction, and monitor the impact of compression and breast thickness on image quality and dose. The 2015 ECR meeting was held March 4-8  in Vienna.

Volpara showcased updates to its suite of breast density and quantitative breast imaging software tools at ECR, including new versions of VolparaDensity for use with digital breast tomosynthesis and VolparaAnalytics, featuring expanded breast imaging metrics and additional quality metrics including positioning and time between study acquisition.  

In the first study, “Effect of volumetric mammographic density on performance of a breast cancer screening program using full-field digital mammography” (C-0360), researchers from UMC Ultrecht and Nijmegen in the Netherlands looked at mammographic screening performance stratified according to volumetric breast density software. Designed to examine to what extent mammographic density affects screening performance, nearly 70,000 mammographic studies were analyzed using VolparaDensity, and recall and breast cancer detection information was obtained from the screening program registration system. The distribution of women in each BI-RADS density category was 19.7% (BI-RADS 1), 43.1% (BI-RADS 2), 29.4% (BI-RADS 3) and 7.7% (BI-RADS 4). For the 421 screen-detected and 150 interval cancers, cancer detection rates were 3.7%, 6.4%, 6.6% and 6.3% in categories 1 through 4, respectively. The interval cancer rate increased as breast density increased, which also resulted in a decrease in sensitivity as density increased, ranging from 85% (BI-RADS 1) to 58.6% (BI-RADS 4) of cancers being detected by mammography.  Additionally, the number of false-positives was also higher with increasing breast density, ranging from 11.4% (BI-RADS 1) to 28.6% (BI-RADS 4).

In a second study, “Impact of quantitative breast density on experienced radiologists’ assessment of mammographic breast density,” (C-1281)  Kathy Schilling, M.D., of Boca Raton Regional Hospital, investigated whether the use of quantitative breast density software improved the consistency in breast density assessment between radiologists. Eight dedicated breast radiologists assessed 88 digital mammographic studies and assigned each study into a BI-RADS density category. After two weeks, the radiologists re-read the studies using VolparaDensity as an interpretive aid. The use of VolparaDensity reduced the variability in the number of women allocated into each density category and significantly improved the inter-observer agreement in radiologists’ assessment of BI-RADS (p=0.0374), with a mean kappa statistic of 0.5664 and 0.6266 without or with VolparaDensity, respectively. Most readers accepted and used the automated scores to improve their readings.

In the third study, “Should volumetric breast density be included in breast cancer prediction models? Proposal of an integrated quantitative and reproducible approach” (B-0238), researchers investigated the relationship between volumetric breast density (VBD) and breast cancer risk estimates using the Tyrer-Cuzick model. For 249 patients undergoing mammographic screening, Tyrer-Cuzick lifetime risk estimates and Volpara Density Grades (VDG) were obtained. Median lifetime risk estimates comparing VDG2 (11.0%) versus VDG3 (14.5%) and VDG4 (15.6%) were significantly different (p=0.0011 and p=0.0002, respectively). Lifetime risks were comparable between VDG3 and VDG4 categories (p=0.0931). Lifetime risk increased with increasing breast density, indicating that VolparaDensity breast density measures could be used with existing risk prediction models to more accurately identify high-risk women.

In a fourth study “Impact of objective volumetric breast density estimates on mean glandular dose calculations in digital mammography” (C-1576), researchers compared mean glandular dose (MGD) estimation using Volpara’s breast glandularity output with the MGD output for 5,076 patients on X-ray systems from various manufacturers. Comparing individual systems, Volpara’s MGD was significantly higher compared to the MGD output by the GE DS, GE Essential, IMS Giotto and one of the Hologic Selenia Dimension. Volpara’s MGD was significantly lower compared to the Philips Microdose and one of the Hologic Selenia Dimension systems. MGD values provided by manufacturers using different models and different assumptions of breast density are significantly different to MGD estimates obtained using the patient’s own breast density.

In the educational exhibit, “Volumetric breast density and BI-RADS 5th edition,” (C-1070) Andre-Robert Grivegnée, M.D., of Institute Jules Bordet, looked at the potential impact of the recent BI-RADS fifth edition updates on the use of VolparaDensity and found: 1) the software can already be configured to output letters (a/b/c/d) to denote the new BI-RADS density categories; 2) removal of the quantitative quartile ranges should not impact visual density assessment; and 3) breasts that are predominantly fatty but still contain regions in the breast that are sufficiently dense to obscure small masses (i.e. focal densities) tend to be scored as BI-RADS c, in accordance with the ACR recommendations. Furthermore, use of the denser breast to assign the final density category (rather than the average) aligns better with the American College of Radiology (ACR) recommendations, and implementing this change into the software would result in <5% of women being re-classified from non-dense (BI-RADS 1 or 2) to dense (BI-RADS 3 or 4). As a result, Grivegnée concluded that volumetric approaches are entirely consistent with the changes in the BI-RADS 5th edition provided that the software is configured to use the denser breast.

For more information: www.volparasolutions.com

Related Content

Lunit Unveiling AI-Based Mammography Solution at RSNA 2018
News | Mammography | November 15, 2018
Medical artificial intelligence (AI) software company Lunit will be returning to the 104th Radiological Society of...
Breast Density Advocate Nancy M. Cappello Passes Away

Nancy Cappello. Image courtesy of AreYouDense.org.

News | Breast Density | November 15, 2018 | Jeff Zagoudis, Associate Editor
Imaging Technology News extends its condolences to the family, friends and colleagues of Nancy M. Cappello, Ph.D., who...
Life Image and Mendel.ai Bringing Artificial Intelligence to Clinical Trial Development
News | Artificial Intelligence | November 15, 2018
Life Image and Mendel.ai announced a new strategic partnership that will facilitate the adoption and enhancement of...
Artificial Intelligence Predicts Alzheimer's Years Before Diagnosis
News | Neuro Imaging | November 14, 2018
Artificial intelligence (AI) technology improves the ability of brain imaging to predict Alzheimer’s disease, according...
Merit Medical Completes Acquisition of Cianna Medical
News | Women's Health | November 14, 2018
Disposable device manufacturer Merit Medical Systems Inc. announced the closing of a definitive merger agreement to...
The MOZART Supra Specimen Tomosynthesis System is the latest generation of 3-D imaging for breast cancer surgery.
News | Breast Imaging | November 08, 2018
KUBTEC announced the launch of a new innovation in the treatment of breast cancer. The Mozart Supra Specimen...
Feature | Breast Imaging | November 07, 2018 | By Jeff Zagoudis
Breast imaging technology has experienced major growth over the last decade or so, and a new report suggests the mark