February 11, 2008

Twin Peeks — Stereo Mammography’s Two Perspectives May Overcome One Mammography Limitation
By Kathy Hardy
Radiology Today
Vol. 9 No. 3 P. 22

Could the technology used to watch a 3D movie provide a better way to view breast tissue and lead to earlier detection of cancerous lesions with fewer false positives?

As researchers describe stereoscopic digital mammography, the basis of this 3D viewing technology is rooted in the same technology that makes objects appear to jump from movie screens or enables View-Master slides to show lifelike views of the Grand Canyon. However, with the advancement of digital imaging, stereoscopic digital mammography has emerged as a 21st century breast imaging tool.

“In a historical context, mammography is not a new idea,” says David J. Getty, PhD, division scientist with stereoscopic mammography developer BBN Technologies. “Radiologists would take two x-ray images separated by a few degrees, but there was no good way to fuse them into a single stereo view. They would put the two images up on a light box, side-by-side, and cross their eyes. When done correctly, you could see the tissue in depth. But this was not a comfortable method of viewing; it caused visual fatigue.

“Now, with the introduction of digital radiography, suddenly there is a way of viewing images that can be done very efficiently with optimum quality,” he adds.

Getty reported results in November at RSNA 2007 showing that stereoscopic digital mammography reduced the number of false positive results by one half in a study of 1,093 patients. Some researchers are optimistic about the value of stereoscopic viewing of breast images in conjunction with such developing modalities as tomosynthesis and CT breast imaging. However, others think stereo is just another potentially confusing way to view digital images.

Old Is New
“The process is not new,” says Carl D’Orsi, MD, director of breast imaging at Emory University’s Winship Cancer Institute in Atlanta. “What’s new is the way we accomplished it. With the development of digital mammography, coupled with a specialized workspace, we’re able to see breast tissue in three dimensions, reducing obstructions that can hide lesions.”

D’Orsi was coauthor of the Emory-conducted study based on a clinical trial of stereoscopic full-field digital mammography compared with standard full-field digital mammography. Both modalities were compared for detection of suspicious breast lesions in the screening process.

“Standard mammography is one of the most difficult radiographic exams to interpret,” Getty says. “In a 2D image of the breast, subtle lesions may be masked by underlying or overlying normal tissue and thus be missed, and normal tissue scattered at different depths can align to mimic a lesion, leading to false positive detections. Stereo avoids that; with stereo, the tissue is distributed in depth and, therefore, is easier to view.”

Getty spent the past 12 of his 30-plus years with BBN working on this technology. An experimental psychologist by training, Getty says he always had an interest in visualization, particularly stereoscopic visualization. His focus at BBN has been in the areas of visual pattern recognition and image-based decision-aiding systems, applications of stereoscopic human vision, and improved human-computer interaction.

“I thought it was strange that radiologists didn’t look to stereo in detecting abnormalities in the breast,” Getty says. “I thought that stereo might improve their capabilities to find them.”

Interim study results were based on data gathered from 1,093 patients determined to be at an elevated risk for developing breast cancer. Each patient received standard digital and stereoscopic digital mammography exams. Through the combined mammography procedures, a total of 259 suspicious findings were detected and referred for additional diagnostic testing. Of those findings, 109 were determined to be true lesions. Comparing the methods of mammography, 40 of the 109 lesions were missed by standard mammography, while the stereoscopic exam failed to detect 24.

When considering the issue of false positives, 150 of the 259 findings were found to be false. Standard mammography yielded 103 false positives; stereo mammography yielded 53. Overall, of those women involved in the study, stereoscopic digital mammography detected more true lesions than standard digital mammography and reduced false-positive findings by 49%.

“This could have a significant impact by cutting in half the number of women who are needlessly recalled for additional diagnostic workups, resulting in a large savings in cost and patient anxiety,” Getty says. “Reducing the number of false positives would have a huge impact.”

The stereoscopic mammography study results showed significant success in finding calcification lesions. Standard mammography missed 20 of 41 calcification lesions, while stereo mammography missed only four of 41 calcification lesions. In this study, stereo successfully reduced the rate of missed calcifications by 80%.

“Stereo mammography is substantially reducing false-negative readings, especially for clustered calcifications,” Getty says. “In 2D, calcifications appear mixed up with other matter within the breast tissue. That is better separated out in stereo.”

“Calcification lesions are the first indicators of the earliest type of breast cancer,” D’Orsi says.

This five-year trial included 1,500 patients at its conclusion in December 2007. Funding for the study came from a grant issued by the U.S. Department of Defense’s congressionally mandated breast and prostate cancer research programs.

The process of acquiring a stereoscopic digital mammogram in some ways varies only slightly from that of a standard mammogram. The breast is held in a fixed position while two digital images are taken using a modified digital mammography unit. Those images are taken from two different points of view, separated by approximately eight degrees.

The viewing is where the new technology makes a real difference. The images are transmitted to a stereo display workstation where they are shown on two high-resolution, crosspolarized LCD monitors attached one on top of the other with a specially coated glass partition between them. A radiologist wearing polarized glasses sees the image on the lower monitor with the left eye while the right eye sees the image on the second monitor reflected from the glass. With this method, the radiologist’s brain is able to fuse the two images to view the internal breast structure in 3D.

Getty says standard digital mammography equipment and software can be transformed to stereo by simply adding a stereo display to existing equipment. He also sees no difficulty on the part of radiology technologists and radiologists in making the transition from 2D standard film or digital mammography to stereoscopic digital mammography. The technologist just needs to take the image pair from different angles. Reading the images should be simple as well, he says.

“Seeing in 3D is a more natural process,” he says. “You’re looking at the same thing you would have looked at with a 2D digital mammogram but in 3D. There’s a ‘wow’ factor, but you get used to that.”

Having the capability to see in stereo, with or without the proper equipment, is key to the ease of using this process. A human’s eyes are separated by approximately 2.5 inches, each with a slightly different view. Stereo vision is the result of those two views fused by the brain, with the object seen in depth. “The brain is able to put together these images taken from different vantage points and figure out where things are in terms of depth,”
Getty says.

However, as D’Orsi points out, “There are some humans who lack stereoscopic vision.”
“But,” Getty adds, “people with visual deficits are unlikely to choose radiology as a profession. I’ve yet to encounter a radiologist who does not have excellent stereoscopic vision.”

Etta Pisano, MD, director of the University of North Carolina Biomedical Research Imaging Center in Chapel Hill, points to the visibility factor as a possible deterrent to radiology professionals adopting stereoscopic mammography, particularly those who have not yet made or are in the process of making the transition from film to digital imaging.

“You still have the same problems as you have with digital,” she says. “You have more data to look at, but with stereoscopic, it’s more complicated. Digital and film look the same, but there are more images and more data. Stereoscopic requires you to look at two images in 3D; some people just can’t do that.”

Pisano, who was a member of the team that developed the first digital mammography device, centers her research on improved technology development for breast cancer diagnosis and the application of study results in current clinical practice. She describes a different modality—digital breast tomosynthesis—as “true 3D” image viewing and sees it as the “next step in technology development beyond digital mammography.” Tomosynthesis, which is currently being studied but has not yet received FDA approval, creates a 3D image of the internal breast structure that eliminates the overlapping of structures within the breast. “The added advantage is that there is no overlapping of 2D images merged into 3D,” she says.

Getty and D’Orsi both contend that stereoscopic mammography could be a first step toward pulling together any “slice-based imaging” such as tomosynthesis. For example, as mammography equipment manufacturers develop machines to perform digital breast tomosynthesis, professionals can see that the tomosynthesis images could be reconstructed and viewed through a stereo mammography system to create a 3D breast image that could be examined from numerous angles, they said.

“With any slice-based technology, you go through slices one by one in a loop mode,” Getty says. “That still raises problems. There’s a chance of missing a cluster of calcifications. There can be a leading piece of calcium in one slice but not in the next two. You may never see more than one or two elements of calcium at one time.”

However, when viewed in stereo, you can “see through the stack and let the radiologist see everything all at once,” Getty says. “You could let them control how much of the stack they’re [seeing] at one time. You could control the width of the slab of slices and move it continually through the breast. The radiologist could also interactively tilt the stack of slices to change the stereo point of view.”

Bringing together stereoscopic and slice-based technologies to create a more realistic view of breast tissue could be the focus of the future, according to D’Orsi.

“I think they would be powerful together,” he says. “One complements the other.”

— Kathy Hardy is a freelance writer and editor based in Phoenixville, Pa.