Breast CT finally enters clinical testing at UC Davis

Breast CT, on the drawing board for decades, is now in phase II clinical testing at the University of California, Davis. Begun in early summer, the second phase of the Breast CT Project will enroll 190 patients over the next few years. As of September, researchers had scanned 17 patients who had been identified through mammography as having BI-RADS 4 or 5 lesions. Fifteen lesions have been confirmed as cancer.

"It's important to use patients with known or suspected breast cancer to ensure that the breast CT scanner is picking it up," said codeveloper John M. Boone, Ph.D., a professor of radiology and biomedical engineering at UC Davis.

Unlike mammography, the breast CT scanner does not require breast compression. Patients lie facedown, and an opening in the table accommodates the breast. A single flat detector mounted on a gantry below the table rotates around each breast. Combined scan time for each breast is 33 seconds. A second prototype machine being built has improved detectors that should cut scan time in half. Researchers ultimately hope to develop a slip ring scanner, further reducing scanning to five seconds.

The breast CT scanner is capable of detecting lesions as small as 4 mm, compared with 11 mm for mammography. So far, Boone and his team, including mammographer Dr. Karen K. Lindfors, chief of breast imaging at UC Davis, have identified microcalcifications, mass lesions, and spiculations. A single scan generates 300 coronal images, and the learning curve for imagers used to viewing mammograms is steep, Boone said.

Images can be viewed in coronal, axial, and sagittal display. Specially developed display software allows visualization of all three views simultaneously. The researchers have also developed software that simulates compression, allowing comparisons with mammograms.

"These are breast images that no one has seen before, at least in the current form," Boone said. "GE built a scanner in 1976, but the technology of the era was fledgling and comparisons are not fair."

In the 1970s, practitioners dismissed breast CT because of concerns about radiation dose and cost-effectiveness. The first thing Boone and codeveloper Thomas Nelson, Ph.D., had to do when they began the project was recalculate the dose. They realized that the older versions used conventional CT scanner technology, which sent x-rays transversely through the thoracic cavity. This exposed nonbreast tissue to radiation and ran the risk of cardiac and respiratory motion artifacts. Their new Monte Carlo estimations put the dose for imaging only the breasts at 4.5 mGy, comparable with two-view mammography (Radiology 2001;221:657-667).

While the team is encouraged by the image quality, they are seeking to improve it. One avenue is to segment glandular tissue from the surrounding fat and other tissues of breast through improved signal-to-noise or contrast-to-noise ratios. But the researchers are inherently at a theoretical disadvantage because they are working in an environment of minimal dose to the breast. This means the data are relatively noisy.

"Our goal is to improve the tissue signal and reduce the noise simultaneously, without throwing out data or otherwise compromising the fidelity of the information," said Nelson, a professor of radiology at the University of California, San Diego.

Scientists at the University of Rochester, University of Massachusetts, and Duke University are also developing breast CT scanners, but the UC Davis effort is the first to have reached clinical testing.

Following a successful phase II trial, the researchers would conduct a larger trial to determine if the new technology can indeed detect tumors earlier than mammography. In the meantime, several other related projects are in the works. The team is developing computer-aided software to work with breast CT, an integrated PET/breast CT system, and an integrated 3D volumetric ultrasound system for use with breast CT.