“Virtual Breast” Designed to Enhance Ultrasound Elastography Cancer Detection

A virtual breast has been developed to help train clinicians in the use of ultrasound elastography. The technique has the potential for improving cancer detection, but only if the results are interpreted accurately.

Women are encouraged to get mammograms to screen for breast cancer, although the tests are imperfect at best. Only a small number of suspicious mammograms actually leads to a cancer diagnosis, which results in needless worry for women and their families, not to mention the time, discomfort, and expense of further testing, including ultrasounds and biopsies. Recently, a different type of test, ultrasound elastography, has been used to pinpoint possible tumors throughout the body, including in the breast.


“It uses imaging to measure the stiffness of tissue, and cancer tissues are stiff,” said Dr. Jingfeng Jiang, a biomedical engineer at Michigan Technological University (Houghton, USA). Those images can be breathtakingly clear: Dr. Jiang reported that in one elastogram, the tumor is as different from normal breast tissue as a yolk is from the white in a fried egg. However, not all images are that precise. “Depending on who does the reading, the accuracy can vary from 95%–40%,” he said. “Forty percent is very bad—you get 50% when you toss a coin. In part, the problem is that ultrasound elastography is a new modality, and people don’t know much about it.”

Ultrasound elastography could become an effective screening approach for women who have suspicious mammograms, but only if the findings are appropriately interpreted. Dr. Jiang, who helped develop ultrasound elastography when he was a postdoctoral researcher at the University of Wisconsin-Madison (USA), reasoned that clinicians might improve their accuracy if they could practice more, and therefore, he and his colleagues set about to construct a virtual breast.

Similar to a simulator used to train fledgling surgeons, their virtual breast—a three-dimensional (3D), computer-generated “phantom”—would let medical professionals practice in the safety of the lab. It was developed using data from the Visible Human Project, which gathered thousands of cross-sectional images from a female cadaver. Therefore, it mimics the complexity of the real thing, integrating a range of tissue types and anatomical structures, such as ligaments and milk ducts.

Clinicians can practice searching for cancer by applying virtual ultrasound elastography to the virtual breast and then evaluating the resulting images. Dr. Jiang hopes that ultimately the software will be available to anyone who needs the training.

The investigators presented a poster on their project at the Institute of Electrical and Electronics Engineers (IEEE) Ultrasonics Symposium, held September 3–6, 2014, in Chicago (IL, USA).

Related Links:
Michigan Technological University