Researchers at Stanford University are developing technologies intended to dramatically improve MRI as a tool for breast imaging while decreasing the false-positive rate and significantly lowering associated medical costs.
Because of its ability to pick up lesions in the breast that mammography may miss, magnetic resonance imaging (MRI) plays an important role in assessing the extent of breast cancer in newly diagnosed women and in monitoring women at high risk for developing the disease. Relying on radio waves created by magnetic fields, MRI creates images without exposing patients to ionizing radiation. Despite these advantages, MRI has limited ability to clearly distinguish a lesion as cancerous or benign. Because of this, breast MRI produces a significant number – about 13 percent or sometimes higher – of false-positive readings, suspicious findings that require further investigation, including biopsy. These workups may lead to patient anxiety and added expense.
Brian Hargreaves, principal investigator and assistant professor of radiology at Stanford, believes that higher resolution scans are the key to improving MRI’s diagnostic capabilities. His research team is developing new hardware and software that will allow radiologists to view key features of suspicious breast lesions such as size, shape, distribution in the breast, and internal architecture, as well as jagged borders that usually indicate malignancy. Current breast MRI scanners lack the clarity to capture those features in fine detail. “We’re trying to improve the overall specificity or accuracy of breast MRI,” says Hargreaves.
Tiny Radio Antennae
To create the sharpest images possible, the researchers built an array of tiny coils that act like antennae, receiving the radio frequency signal produced by the body when a magnetic field is applied. Image acquisition is faster with the coil array because the individual coils pick up the signal from different parts of the body simultaneously.
Determining the correct coil size was critical. As Hargreaves explains, a coil that is too small will not pick up the signal from inside the body, and coils that are too large pick up too much interference or noise. Graduate student Anderson Nnewihe built an 18-coil array “bra” that flanks both breasts in a conventional breast MRI scanner. Lying face down, patients place their breasts in the bowl-shaped holders and scanning begins.
Hargreaves notes that breast MRI presents a special challenge to image acquisition time. Breast MRI usually requires intravenous injection of an imaging agent that is taken up by the tumor in 30 to 60 seconds and infiltrates surrounding tissue within 5 to 8 minutes, often obscuring the tumor. “You want to image quickly to see which areas light up first,” says Hargreaves.
To create sharp images quickly, Kyung Sung, a researcher working with Hargreaves, has developed a process that does more with less information. A conventional breast MRI scan takes up to an hour because of the large amount of data collected to create a scan. “We collect little bits of information in a special way that enables us to make images as though we had collected more details at each point,” Hargreaves explains. Using specially designed software, a computer “builds” the final image from a series of sparse signals collected from the body. Each successive signal layer fills in the final image with help from the computer program. The result is a sharp image created with a fraction of the data used for conventional breast MRI scans. The Stanford process is similar to filming a movie with a very fast, but low-pixel camera. The computer program then transforms the movie so that it is indistinguishable from one taken with a slower, high pixel or “HD” camera, but acquired at the much higher speed. “The more we know about enhancement patterns in normal tissue and tumors, the more we can use this information to sharpen the image,” Hargreaves says.
Enhancing MRI’s ability to detect breast lesions also requires integrating techniques to suppress fat within the breast. Fat obscures diagnosis because it shows up very brightly on most MRI sequences. Because the MRI signal of fat has a unique frequency like a radio station, the researchers use a technique to tune out the fat’s signal.
Magnifying Milk Ducts
The improved sharpness provided by the coil array allows researchers to image submillimeter milk ducts where most breast cancers begin. In a nonlactating woman, milk ducts collapse. These structures sometimes become filled with rogue cells that can lead to benign and malignant conditions such as intraductal papilloma and ductal carcinoma in situ. The Stanford approach is “like putting a magnifying glass in or around the duct,” says Stanford radiologist Bruce Daniel who is providing clinical guidance on the project. Using clinical findings obtained through surgery or biopsy, the researchers can correlate the appearance of the duct or the pattern of the cells found on the MRI scans.
As the researchers introduce the new breast coils and image acquisition techniques into the clinic, they will assess whether they can see tumors more clearly, distinguish tumors from similar benign conditions, and observe which breast area images are made sharper by the new approach. They will also determine whether the techniques reveal important new features not previously visible. Their findings may eventually extend the power of MRI to identify very small breast cancers on the first pass. Hargreaves anticipates that the new technique may reveal tumors as small as 2 mm – one-half the size of tumors picked up by conventional breast MRI.
Making MRI an Option for More Women
Moving the Stanford approach beyond the lab will require the alignment of several key activities. Commercialization of the coil arrays is critical, and Hargreaves notes that MRI manufacturer General Electric (GE) has assisted with and licensed the Stanford breast coil technology. Large clinical studies are also needed to determine whether the Stanford approach provides improved sensitivity to detect cancerous lesions.
By improving breast MRI image resolution, Daniel sees increased opportunities for MRI among patients undergoing chemotherapy. “The ways in which we treat breast cancer are exploding. MRI could be used to inform clinicians on how patients’ individual tumors are responding to treatment,” he says.
Three-Dimensional MRI Using Water Diffusion
Daniel and Hargreaves also are collaborating on another project that, if successful, could eliminate the need for intravenous breast MRI imaging agents in some cases. The technique creates high-resolution three-dimensional (3D) MR images using diffusion (or spreading) of water molecules to reveal tumors. Water diffusing through normal, healthy tissue appears dark, but when the water molecules encounter densely packed tumor cells, diffusion is restricted making the tumor appear brighter than other tissue. By eliminating the imaging agent, the cost per patient study drops substantially, a technician can perform the scan, and the entire process is pain-free.
Positive results from all of these efforts may make breast MRI an option for many women who otherwise would not be candidates for the technology according to current guidelines. Hargreaves and Daniel note that the imaging techniques will also be applied to visualization of cancers in other organ systems and that the 3D diffusion imaging technique shows promise for imaging of cartilage in joints and between discs in the spine.
This work was supported in part by the National Institute of Biomedical Imaging and Bioengineering.
Nnewihe AN, Grafendorfer T, Daniel BL, Calderon P, Alley MT, Robb F, Hargreaves BA. Custom-fitted 16-channel bilateral breast coil for bidirectional parallel imaging. Magnetic Resonance in Medicine. Wiley-Liss Inc. First published online 2011 Feb 1.