Q&A: Evaluating Breast Density

May 22, 2012

Ralph Highnam, CEO of technology company Matakina International, weighs in on the challenges and promises of evaluating objective breast density.

Ralph Highnam, CEO of technology company Matakina International, weighs in on the challenges and promises of evaluating objective breast density.

Why has there been so much attention on standardizing the measurement of breast density?

Linked to an increase in breast cancer risk and the decreased sensitivity of mammography, breast density has received a great deal of attention lately in the national consumer media, as well as within the imaging community. The need for accurate, reproducible density measurement was referenced at a recent FDA panel meeting for Automated Breast Ultrasound and in the ACR’s recent statement on breast density.

Breast density is the volume of dense tissue divided by the volume of the breast and then multiplied by 100 to get a percentage. Density has historically been measured by radiologists comparing the light and dark parts of a mammogram, and is reported using the BI-RADS density categories.

While this provides a relative picture of density, this subjective density assessment is operator dependent and suffers from significant detractions. First, it is entirely qualitative, with agreement between interpreting physicians in estimating density by quartile at only about 60 percent in the middle two BI-RADS density categories, in which the differentiating line between low and high tissue density is found, according to a 2006 study in Academic Radiology.

The second challenge of the BI-RADS reporting system for tissue density is that it is two dimensional and the breast is a three dimensional organ that varies dramatically in shape, size and composition between individuals. An area that appears almost white on a mammogram could be a single highly dense area or it could be several densities overlying each other. Density could be evenly distributed in the breast, or the area near the skin could be dense with the center of the breast being largely replaced by fatty tissue. It is impossible to tell from simply viewing the two dimensional images.

Our software, Volpara, for example, determines density from the information in the raw digital mammogram itself. That information does not vary, so the next phase of development involves comparing density over time, which we believe could help detect cancers earlier by picking up the first signs of change.

How do you evaluate volumetric density from 2-D images?

Mammograms are 2-D images, but each pixel in that 2-D image represents a measure of the X-ray attenuation between that pixel and the X-ray source. Knowing the X-ray attenuation allows you to determine the volumes of the various tissue types.

Volpara provides an objective volumetric assessment of breast tissue density. Using the “for processing” digital images from a mammogram, and a state-of-the-art algorithm developed by some of the world’s leading imaging scientists, the software presents interpreting physicians with an assessment of the percentage of dense tissue contained within the breast and a mapping of that percentage to a Volpara Density Grade (VDG). The VDG is a number from 1 to 4 and the mapping has been set so as to optimize the relationship between VDG and BI-RADS breast density category.

Volumetric breast tissue density is defined as the volume of fibroglandular tissue divided by the total volume of tissue within the breast. Using segmentation routines, the pectoral muscle is removed to help get an accurate volume. Next, the skin is excluded, because breast cancers occur in fibroglandular tissue, not skin and, likewise, breast fat cells do not become cancerous.

The breast tissue density equation can then be written: Volumetric Breast Density = Fibroglandular Tissue / (Fibroglandular Tissue + Fat)

Determining volumetric density involves knowing the volume of the breast and defining the amount of fat within that volume. Within the image, the critical factor is being able to define a grayscale value for fat. Using proprietary algorithms, we have been able to determine the unique “fat value” for each patient’s mammogram, subtract the volume of fat from the total volume of the breast to provide the volume of fibroglandular tissue.

Applicable to both clinical and research applications, it is important to realize that because a breast density measurement using VDG is always based on the fat value for an individual patient’s mammogram, the value will only change with actual changes in the breast composition and not due to different imaging physics or different processing algorithms getting the images ready for display. This allows the breast density values from one year to be compared with other years, and values between populations to be compared one to the other as well as temporally.

How does an objective density measurement correlate to the BI-RADS scale?

Volumetric density is always significantly lower than density measured in 2D. The reason is that most of every breast is fat, even in apparently very dense individuals. When one looks at a 2D image, however, what you are seeing are overlapping, projected areas of density, which tends to exaggerate the overall density by a factor of roughly 3X. Despite that difference in scaling, we do however find that Volpara correlates very well with BI-RADS density findings from MQSA qualified radiologists and thus we’ve been able to also output a VDG which provides a surrogate for the BI-RADS density assessment.

How would this objective measure be used to make imaging decisions?

Globally, many different uses are being studied by researchers ranging from choice of screening imaging modality to adjustment of screening intervals.

As an example from the U.S., if a practice determined that all patients with high density (BI-RADS 3-4) were to be offered supplementary screening ultrasound, this decision could be put in the form of an operational protocol for care, allowing the mammography technologist, who has the VDG assessment available on her workstation, to offer such patients the ultrasound procedure following the screening mammogram. Studies have shown that convenience is a major factor in women’s care, and being able to provide the supplemental ultrasound exam as a part of the screening visit increases substantially the number of women who elect the procedure.