News Story

Running the Numbers

54% more likely to develop breast cancer were women identified as BRCA 1 or 2 carriers who had ever had a chest x-ray than women who had never undergone the procedure, according to a recent study.¹ The International BRCA 1/2 Carrier Cohort Study, conducted by a consortium of European cancer centers, analyzed questionnaires from 1,600 women who carry BRCA 1/2 mutations. Researchers also found that women who were exposed to x-rays before age 20 had a 2.5-fold increased risk of developing the disease before age 40, compared with women who had never been exposed. “Since BRCA proteins are integral in repairing damage to breast cells, we hypothesized that women with BRCA 1/2 mutations would be less able to repair damage caused to DNA by ionizing radiation,” said David E. Goldgar, PhD, a lead author of the study who was the chief of the Genetic Epidemiology Group at the International Agency for Research on Cancer (Lyon, France) at the time the research was conducted. “Our findings support this hypothesis and stress the need for prospective studies.”

Reference

1. Andrieu N, Easton DF, Chang-Claude J, et al. Effect of chest x-rays on the risk of breast cancer among BRCA 1/2 mutation carriers in the international BRCA 1/2 carrier cohort study: a report from the EMBRACE, GENEPSO, GEO-HEBON, and IBCCS Collaborators’ Group. J Clin Onc. 2006;24:3361?3366. Available at: http://www.jco.org/cgi/content/abstract/24/21/3361. Accessed September 8, 2006.

KB-CAD System Provides “Search Engine” for Mammography

A new system developed by medical physicists at Duke University (Durham, NC) can speed up the process of computer-aided detection (CAD) using a search-engine-like query of a mammogram database. The system, known as knowledge-based computer-assisted detection (KB-CAD), works by comparing mammograms to images of known breast cancer using the principles of information theory for better speed and efficiency.

“The user starts by clicking on an image location that appears suspicious,” explains Georgia Tourassi, PhD, associate research professor in the department of radiology at the Duke University School of Medicine, who presented the system at the 48th Annual Meeting of the American Association of Physicists in Medicine (AAPM) in Orlando, Fla. “The system compares the image neighborhood around the query location with images stored in the knowledge database. Our KB-CAD system is intended as an interactive computer aid to help the user analyze any mammographic location he or she wants.”

KB-CAD differs from ordinary CAD, however, in that it maximizes system performance by comparing the unknown case with the most informative stored cases. Tradi-tional CAD systems are weighed down with comparing the mammogram in question to every image in the database; however, KB-CAD indexes its cases based on image entropy.

“Entropy is used to quantify image complexity by looking at the distribution of the gray level values of the image pixels,” Tourassi explains. “Images with greatly varied distributions of image pixel values have higher entropy than images with more uniform distributions. The normal breast parenchyma can be as complex as abnormal tissue in mammograms, often resulting in similar entropies.

“In our preliminary studies,” she continues, “we observed that knowledge cases with higher entropy—in other words, higher complexity—are more useful to the system in terms of making better decisions. This result suggests that instead of blindly depositing mammograms in the knowledge database, we may need to consider a more selective deposit scheme where only complex mammograms with higher entropy are deposited.”

For their pilot study, Duke researchers debuted KB-CAD on a database of 2,300 images. Each sample image was compared to only the 600 most informative stored cases as determined by image entropy, which cut analysis time to less than 3 seconds per search. “Entropy calculations are very fast,” Tourassi says. “The calculation of mutual information is computationally very expensive. Thus, by following the entropy-based indexing scheme, we can substantially cut down on the computational demands of the KB-CAD system without sacrificing diagnostic performance.” Tourassi says there are no particular requirements for the technology used to implement the system.

The researchers have now turned their attention to optimizing the system parameters, studying the clinical impact of the interactive paradigm, and developing a system that will easily translate across a variety of platforms.

—C. Vasko

Product Showcase: Skin Markers Geared Toward Women With Sensitive Skin

Beekley?s Soft ?n? Stretchy SPOTS mark scars, moles, and nipples during mammography exams.

Beekley Corp (Bristol, Conn) recently introduced the latest addition to its line of mammography skin markers. The new Soft ‘n’ Stretchy SPOTS, which mark scars, moles, and nipples, are designed to stick better and peel away with less resistance, resulting in a more comfortable patient experience.

The new markers are intended to reduce the need for retakes by sticking better during positioning and compression; the material also bends and expands with the patient’s breast during compression, giving technologists the opportunity to focus more on completing the scan with all possible expedience. When the exam is finished, two tabs are pulled in opposite directions, allowing the sticker to be lifted away quickly and painlessly.

Soft ‘n’ Stretchy SPOTS come in four shapes for unmistakable images on film, and they are available in two designs: pastel pink or floral. For more information or to request samples, visit www.beekley.com.