X-ray mammography has long been regarded as the trusted workhorse of breast radiology. Practices may have switched from film-based image capture to digital imaging, but the basics of the examination have changed little. Each breast is compressed in turn while a low dose of x-rays is directed through the tissue from two orthogonal directions (top-to-bottom and/or side-to-side). The end result is a collection of 2D images; one per breast for single-view mammography, two per breast for double-view.

This perception of x-ray breast imaging as an exclusively 2D technology is now beginning to change. The appearance of digital tomosynthesis has shown that mammography-style images can be acquired slice by slice across the whole breast volume, providing reporting radiologists with a stack of thin-slice views that can be viewed individually or in cine mode. This method of data capture should, in theory, make it easier to spot tiny lesions or signs of disease that might be masked by overlapping structures on conventional 2D mammograms.

Several vendors with existing interests in conventional mammography are now investing in digital tomosynthesis. None of the systems are yet available commercially in the U.S., Canada, or China, though the U.S. Food and Drug Administration (FDA) has permitted investigational studies to be conducted. In Europe, where digital tomosynthesis is approved for clinical use, early adopters are continuing to probe the technology’s impact on diagnosis and patient management.

The method of data capture is essentially the same, regardless of which vendor’s system is selected: multiple projection images are acquired while the x-ray tube sweeps around a portion of the compressed breast. The angular range of the x-ray tube may, however, depend on the manufacturer and/or departmental preference. Radiologists at Malmö University Hospital (UMAS) in Sweden, for example, are using a wide scan angle of 40º and taking 25 pictures, a process that lasts 20 seconds. Other groups are going with a much smaller scan angle of 15º and taking just nine or 10 images, cutting the time for data acquisition down to 10 or 15 seconds.

“The argument for doing one or the other always comes down to speed,” said Dr. Ingvar Andersson, chief radiologist at UMAS. “You want to make a quick examination, but you still want good images. We have found, for instance, that the wider the scan angle, the more projection images you make, the better the images. But it takes more time.”

Work conducted at UMAS has already shown that 3D tomosynthesis is superior to conventional 2D mammography for the detection of small cancers, particularly when lesions are spiculated. Small cancers can be missed on 2D mammography owing to the superposition of glandular tissue, Andersson said. Some cancers, however, grow diffusely rather than forming a typical tumor-like mass, creating only slight architectural distortion on mammograms. These could still be hard to find on tomosynthesis.

Initial experience at Leuven University Hospital in Belgium similarly suggests that the predicted benefits of 3D tomosynthesis will primarily apply to spiculated lesions. More work is needed to determine if tomosynthesis improves the detection of diffuse lesions, such as invasive lobular carcinoma, said Dr. Chantal Van Ongeval, radiologist at Leuven University Hospital. Tomosynthesis is also unlikely to make it much easier to find small mass-like lesions in dense breast tissue, she said.

PRACTICE MATTERS

So what difference could tomosynthesis make to women with latent breast disease or symptoms of possible cancer? Could this up-and-coming modality make a positive difference to their management? That is the next question for radiology researchers to address, according to Andersson.

“We have shown that it is clearly superior, statistically significantly better in detection as well as the staging of breast cancer, in our selected series of patients. But the question is: what does that mean in an unselected, larger population, like a screening population? What does it mean in practical terms?” he said.

UMAS researchers aim to answer this question through a population-based single-arm study involving 15,000 screening candidates who will be examined with 2D mammography as well as tomosynthesis. The images will be read separately as well as together to gauge sensitivity and specificity, and the characteristics of any extra cancers detected will be analyzed. A cost-benefit analysis will be performed.

Radiologists must also become familiar with the presentation of normal breast structures on tomosynthesis, Van Ongeval said. The removal of superposed anatomy from the thin-slice images has revealed more of everything, not just more malignancies.