Mammography shifts toward spectral imaging

November 10, 2009

X-ray imaging has provided important diagnostic information for clinical applications for almost a century.

X-ray imaging has provided important diagnostic information for clinical applications for almost a century. The sight of a medical doctor holding up a black and- white radiograph against the light is common in TV programs and films.

It is well known that variations in x-ray attenuation create different gray-scale levels that reveal detailed structural information with high spatial resolution. This remains true so long as long as the object's components have differing x-ray absorption properties.

X-rays have a spectrum of wavelengths that correspond to different energies and different colors, just as visible light does. Colors in visible light can help us to detect details in our environment. The same color information from x-rays might be able to help improve diagnostic imaging. The only problem is that detector technology has been unable to resolve the different colors of the x-ray spectrum.

The detection of color requires that each x-ray photon be counted and processed individually and its energy measured (Figure 1). High-speed x-ray counting is now possible, thanks to developments in CT detector technology. The first color CT images were displayed at the November 2008 meeting of the Radiological Society of North America (Figure 2).1

The prospect of color imaging with x-rays has spurred substantial activity among the major medical imaging vendors.2-4 Additional data provided by color for different elements in an object greatly increases the information content of images.5 This could be used to improve visual discrimination between bone and soft tissue and between atherosclerotic plaque and calcifications, or to reduce the required concentration of injected contrast media (Figure 3).6-8

LOW DOSE, HIGH SIGNAL-TO-NOISE

Photon counting has already gained wide clinical acceptance in mammography, and systems incorporating this technology have been available commercially for a few years. Mammography is one of the most common radiographic procedures. More than 100 million women worldwide are screened every year for early signs of breast cancer and this has proven to be the most effective way to reduce breast cancer's mortality rate.9

X-ray mammography does, nonetheless, expose women to ionizing radiation. Although the risk to an individual woman from a screening scan is extremely small, so many healthy women are examined that even a small risk becomes important when it is multiplied by a large number of examinations. The radiation dose from mammography should, therefore, be kept as low as possible.10

Photon-counting detectors can provide high-quality images with extremely low doses of radiation (Figure 4). This is made possible by their ability to eliminate all electronic noise. The signal in ordinary x-ray detectors often “drowns” in noise if the radiation is reduced below a certain level. In other words, the sensor is no longer “quantum limited.” 11 This relationship is analogous to the challenge of hearing in a noisy room. Speaking louder (increasing the dose) improves hearing, but if ambient noise is reduced, then a whisper is loud enough. The efficiency of photoncounting detectors is constant down to zero radiation dose (Figure 5).11

Dose reduction has been the main reason for incorporating photon- counting technology into radiological equipment. A single energy threshold is sufficient to achieve this reduction.

The value of spectral imaging in mammography is now being investigated in clinical trials in several European countries.12 Spectral imaging of the breast may improve the resolution of glandular and adipose tissue and microcalcifications. Radiologists may also be able to take advantage of the fact that elements with higher atomic numbers accumulate in the cancerous breast.13,14 These different elements could be colorcoded according to their atomic number for spectral imaging. This could potentially simplify the processes of disease detection and diagnosis.

CONTRAST MAMMOGRAPHY

Iodinated contrast agents are not used in x-ray mammography, although they are commonplace in many other applications, such as angiography. Gadolinium-based contrast agents are used in breast MRI. Blood vessels resulting from neo-angiogenesis in the cancer are of low quality and leaky. This makes them stand out on Gd-enhanced MRI because the contrast gets trapped interstitially.

The main drawbacks of breast MRI are its low specificity compared with mammography and the relatively high cost per examination. Breast MRI has, nonetheless, come to be recognized as an important tool for the diagnosis of breast cancer and for treatment monitoring.

For example, MRI can be used to check whether neoadjuvant chemotherapy is causing a breast tumor to shrink, as intended. In addition, the rate of neovascularization in a primary breast tumor is directly related to its ability to metastasize. Imaging the vascular structure is, consequently, extremely important.

Several groups have now presented promising results from proof-of-principle “contrast mammography” studies.15-17 The idea here is to use intravenous iodine to reveal the vascular structure in a similar way to the Gd-based agents used in breast MRI. This method could potentially overcome some of the drawbacks associated with breast MRI, such as time and cost.

Different methods have been proposed for performing contrast mammography. One way to visualize the iodine is to subtract a precontrast image from a postcontrast image (Figure 6).15,16

An alternative approach is to use dual-energy imaging, in which two consecutive images are registered at different energies, 30 kVp and 44 kVp, for instance.17 An 8-mm-thick aluminum filter is used to absorb low-energy x-rays, ensuring that most of the x-ray spectrum is above iodine's K-absorption edge (33.2 keV). X-rays are filtered before they enter the breast.

To perform the mathematical subtraction, it is optimum to have about the same number of x-rays above and below the K-edge of iodine. The x-rays with energy below the K-edge will form the low-energy image, and the x-rays with energy above the K-edge will form the high-energy image. Both images need a sufficient quantity of x-rays in order to perform the subtraction, as the two images are subtracted to isolate the iodine signal (Figure 7).

The dual-energy approach to contrast mammography leads to fewer motion artifacts than pre- and postcontrast image subtraction because the time interval between the two images is reduced. In this case, you do not have to wait after you have taken the precontrast image for the injected iodine to reach the breasts in order to take the postcontrast image; the two images taken at different energies are enough to visualize the iodine. Acquiring a series of dual-energy images over time can also reveal information on contrast kinetics, such as uptake and washout, which may yield additional diagnostic information.

Contrast mammography is a promising application for spectral x-ray imaging. The optimum kVp for imaging the breast is much lower than it is for whole-body imaging. The x-ray spectrum can consequently be split exactly at the K-edge, creating high- and lowenergy images simultaneously. This can be done electronically using a spectral sensor that measures x-ray energy.

The two main advantages of this approach to contrast mammography are improved image quality and ease of image acquisition. The image is acquired in the same manner as standard x-ray mammography, so there is no extra work for the radiographer. Motion blurring is eliminated entirely because the high- and low-energy images are acquired at the same time. Use of the photon-counting sensor means that the division in energies is more or less exact, and, unlike traditional dualenergy imaging techniques, the overlap is negligible. Any overlap between the high- and low-energy spectra can reduce the signal-to-noise ratio.6

Spectral imaging of the breast using iodinated contrast has been evaluated in phantom studies.6,15 The technique is now being compared with breast MRI in clinical trials.12

Neither contrast mammography nor breast MRI is likely to be accepted as a population-based screening test given the need for contrast injection. Both technologies may, however, be appropriate for screening patients at high risk of developing breast cancer owing to genetic deficiencies or a family history of breast cancer.

The HighReX Consortium, which is led by Sectra and includes 13 clinical and industrial partners from six European countries, is planning to evaluate the potential for contrast-free spectral breast imaging in a new clinical trial.12 The consortium hopes that the identification of different elements in the breast, such as adipose and glandular tissue, microcalcifications, and cysts, as well as the detection of differences in x-ray attenuation, will increase specificity and sensitivity.

POTENTIAL FOR GOOD?

Photon counting has several clear advantages from a physics point of view. The bottom line is whether the benefits it brings to clinical radiology outweigh the problems. One argument against the development of photon counting, for example, is that some x-ray detector systems have such low noise that further reductions will add little to diagnostic accuracy.

But it could equally be argued that sensors will always vary in performance and the elimination of electronic noise is always an advantage. Noise may vary with time and can be triggered by external sources, such as other instruments or medical devices. The trend toward lower dose imaging also increases the relative importance of reducing electronic noise.

A color image can be obtained from differently colored incident x-rays through repetitive exposures with different kVp values, through dual-source imaging at different energies, or by fastswitching the kVp energy from a single source. Why complicate things with photon counting when this technology is already available in commercial CT systems? After all, these techniques have demonstrated considerable clinical value.18

An argument can also be made, however, in support of the clinical benefits of photon counting, which will take spectral imaging one step further.

Photon counting requires a single exposure and may be less expensive than technologies that use multiple detectors and sources. The technique has been proven in mammography, and once the sensor technology is mature enough, there are no broad reasons why it should not be applied to general xray imaging applications. Conceptually, photon counting is the intuitive way to detect x-rays, which, by nature, are digital and have a color spectra.