banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Volumetric breast density evaluation from ultrasound tomography images
Rent this article for


Image of FIG. 1.
FIG. 1.

Left: the breast phantom provides tissue-equivalent scanning characteristics with simulated fat inclusions embedded in fibroglandular tissue. Right: close up of the phantom surrounded by the transducer ring that translates in the z-direction to image the entire phantom volume.

Image of FIG. 2.
FIG. 2.

(a) Phantom cross section as provided by the manufacturer, where F denotes “fat” and the remainder is fibroglandular tissue (Ref. 46), (b) a sound speed tomogram (top) and the resulting k-means clustering segmentation of the inclusions from the sound speed scan (bottom), and (c) a CT tomogram (top) with its corresponding segmentation (bottom).

Image of FIG. 3.
FIG. 3.

A comparison of the integrated area for the segmented CT scan and the segmented sound speed scan the CURE phantom. The sound speed scans underestimated both large fat inclusions, which may be due to partial volume effects.

Image of FIG. 4.
FIG. 4.

(Left) Sound speed tomogram (scale is sound speed in m/s) and (right) k-means clustered sound speed image (scale is cluster number). Note the high sound speed regions are clearly demarcated by using the clustering technique.

Image of FIG. 5.
FIG. 5.

A comparison of using every slice and every fifth slice for the USPD calculation in vivo. The strong correlation (Pearson ) between techniques justified the reduction of slices used in the USPD analysis.

Image of FIG. 6.
FIG. 6.

Boxplot showing the strong correlation [Spearman ] between the USPD for 93 patients categorized by BI-RADS compositional category. The differences between all BI-RADS categories were found to be significant using a one-way ANOVA and Scheffé posthoc analysis, demonstrating the agreement of the USPD technique with the current standard of care.

Image of FIG. 7.
FIG. 7.

A comparison of the ultrasound percent density (USPD) for every fifth sound speed tomogram and quantitative mammographic percent density (MPD) for 93 patients. (a) Observed data for the CC view, (b) volume to area fit, Pearson , (c) observed data for the MLO view, (d) volume to area fit, Pearson .

Image of FIG. 8.
FIG. 8.

Sound speed images of a BI-RADS category 3 (heterogeneous) breast reconstructed in the sagittal plane (left to right), with every fifth slice shown (legend shows sound speed in m/s). Being able to investigate different cross sections of the breast eliminates the effects that surrounding tissue may have on decreasing image contrast, and allows a more ready comparison to conventional mammography.


Generic image for table

The BI-RADS compositional category distribution for the patient population.

Generic image for table

Mean integrated areas and statistics for each phantom component segmented from four different CURE sound speed and CT scans.

Generic image for table

The mean USPD, standard deviation of the mean, and standard error for each BI-RADS category.

Generic image for table

An analysis of the relationship between volumetric ultrasound percent density from CURE sound speed tomograms and mammographic percent density for CC and MLO views obtained by a volume to area comparison model.


Article metrics loading...


Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Volumetric breast density evaluation from ultrasound tomography images