08 Mar 2007
Mammography is currently the gold standard in breast imaging, and it is used to screen millions of women each year for breast cancer. It has its limitations, however, and lately there has been much interest in emerging breast-imaging technologies. Now, a group from the University of Manitoba in Winnipeg, Canada, has come up with an idea that could significantly improve the sensitivity of X-ray-based breast-imaging modalities, such as traditional mammography or CT mammography, without increasing the radiation dose (Medical Physics 34 256).
Conventional mammography uses a beam of X-rays to image the compressed breast from one direction. It is a cheap, simple and (in most cases) highly effective technique. The drawback is that mammography has trouble detecting abnormalities in women with dense breasts or implants, so misses about 15% of cancers. "Many experts in the field feel there is room to improve the sensitivity and specificity [of mammography]," said Eric Van Uytven, a physicist at Manitoba.
As such, a host of new breast-imaging technologies have been hitting the journals lately. These range from X-ray techniques, such as breast CT and tomosynthesis, to the application of costlier modalities, such as PET and MRI (see Special report: breast imaging on medicalphysicsweb).
The problem with X-rays is that they are highly scattered by human tissue, and this effect greatly reduces the contrast of the resulting images if it is not accounted for. Obtaining a usable mammogram generally means a higher than necessary radiation dose, which is undesirable because of the potential for radiation-induced cancers later on. So radiographers are left with a compromise: a low-dose system that works fine in the majority of women but performs badly when it comes to dense breasts and very small tumours. This is where the Manitoba group's technique comes in.
Van Uytven and colleagues realized that, with the latest energy-sensitive detectors, they should be able to extract information from the scattered photons and use this to increase the sensitivity of mammography and breast CT. "We developed Compton-coherent scatter radiography (CCSR) with the aim of utilizing the scattered X-rays present in any X-ray modality to enhance the information content of the resulting image with no increase in dose," he explained.
CCSR exploits an array of energy-sensitive detectors and an algorithm that can reconstruct the details of the scattering medium from the spatial and energy distribution of the detected photons. The team claims that CCSR can obtain 3D information about an object from a single beam projection (as used in conventional mammography). What's more, the approach should work with existing imaging systems. "CCSR is cost-effective because the only equipment necessary in addition to the parent modality, is an energy-sensitive flat-panel detector," said Van Uytven.
So far the team has tried out the algorithm on a simulated system consisting of a polyenergetic pencil X-ray beam, a breast-tissue phantom containing 0.125 mm3 calcifications, and an array of detectors with resolution similar to currently available semiconductor spectrometers. "Preliminary results indicate that the information content in Compton scatter is sufficient to detect small simulated calcifications to within 0.25 cm in an 8 cm breast phantom," the team noted in Medical Physics.
The researchers are now designing a prototype system that will probably consist of a standard mammographic X-ray tube, a phantom and phantom support, and a scanning-point energy-sensitive detector. They are also working on a benchtop CT system to investigate the benefits of enhancing breast CT with CCSR. "Several challenges need to be overcome, and we are still in the design stage. However, we feel that CCSR is a promising technique that will help existing and novel X-ray breast-imaging techniques convert X-ray scatter from a limitation to a benefit," Van Uytven concluded.