All of the preceding systems operate on the principle of accumulating the signal from all of the x-rays that fall on a del during a particular exposure and digitizing this to form an element of the image. It is also possible to build x-ray detectors in which each interacting x-ray quantum is counted individually.
This approach has a few advantages. Previously, the effect on noise because of the variation in signal produced by an interacting x-ray was discussed. If, instead, the detec- tor simply registers a count whenever an x-ray interacts, then no noise is associated with energy conversion, only the primary fluctuation of the number of quanta interacting with the detector. In addition, an x-ray produces exactly one count regardless of its energy. Some authors have argued that the standard x-ray detector places a higher weight on the higher energy x-rays that interact, because these produce more light, but less contrast. Quantum counting removes this weighting.
The greatest challenge for quantum counting systems is to be able to accommodate the high rate of x-rays interact- ing per second in each del. For the unattenuated beam, this can be 106 quanta/s or higher. Modern electronics can now handle these rates. In the future, it may also be feasible to take this approach another step forward and analyze the energy carried by each interacting quantum and weight the signals thus produced to provide optimum information.
Currently two quantum counting systems are under development. Both are based on the principle of a linear detector; however, to acquire an image in a reasonably short time, both systems use multiple line detectors and move
them during acquisition to fully cover the image plane. This necessitates extremely precise motion control to avoid gaps or overlap when stitching the image together.
The first system, by Sectra (Stockholm) employs crys- talline silicon crystals as direct x-ray absorbers (Fig. 3-11).
The charge produced from each interacting x-ray is col- lected in an electric field and shaped into a pulse, which is counted. In the second system, built by XCounter (Stock- holm), the x-rays interact with a high pressure gas inside the detector vessel and the ions produced are used to form the pulse (Fig. 3-12).
SPATIAL RESOLUTION
The spatial resolution characteristics of each detector type are somewhat different. In the CsI(Tl) systems, light pro- duced in the phosphor attempts to spread isotropically from the point of emission. The CsI crystals tend to channel the light down the length of the crystal by total internal reflec- tion and to some extent this mitigates the spread of light.
Nevertheless, as the detector is made thicker to increase η, there is more spreading of light and a decrease in spatial resolution.
In the CR system, the spatial localization is determined by the laser and, therefore, the spread of light emitted from the phosphor is not important. On the other hand, the laser light that scans the plate scatters on entering the phosphor material causing traps laterally displaced from the point of incidence to be discharged and to contribute to the signal.
This can have a marked effect in reducing resolution and 24 Digital Mammography
FIGURE 3-11. X-ray counting detector based on silicon. (Photograph courtesy Sectra, Stock- holm.)
the effect becomes increasingly important as the thickness of the plate increases. Therefore, with these systems there is a trade-off between ηand spatial resolution.
This issue becomes much less important in the direct con- version detector. The charge signal is moved quickly toward the collection electrode by the electric field before the charge has much opportunity to spread. Therefore, by appropriate choice of electric field, the resolution can be maintained at a high level while at the same time keeping the detector thick- ness large enough to ensure high resolution. Sample MTFs of these systems are presented in Figure 3-13. Because factors other than the del aperture, d, affect the MTF, the ordering of the curves does not necessarily correlate with the del sizes.
AUTOMATIC EXPOSURE CONTROL
Digital image acquisition provides enormous opportunities for automatic optimization of image acquisition. For exam- ple, it is no longer necessary to have a separate AEC sensor as part of the system because the digital detector can serve as a multielement sensor. To use the detector in this way, it is nec- essary that the detector can be read out quickly to determine what the optimum exposure factors should be. This may impose difficulty for some of the current detectors and cre- ative ways of accomplishing this will have to be found.
Optimization of exposure can consider various statistics from a short test pulse made prior to the actual image Detectors for Digital Mammography 25
FIGURE 3-13. MTFs of some current digital mammog- raphy systems.
FIGURE 3-12. High pressure gaseous x-ray counting detector. (Photo courtesy Xcounter, Stock- holm.)
acquisition. These might include the minimum signal from the most attenuating region of the breast. The algorithm can require that in the actual exposure, this signal be greater than some preset value. Alternatively, the SDNR can be predicted, and the exposure can be planned to achieve no less than a certain SDNR in any part of the image. Addi- tional data on the compression thickness and force can also be used to refine the algorithm. These opportunities are only beginning to be realized today and exploiting them
remains an exciting challenge for researchers and system designers.
SUGGESTED READINGS
1. Swank RK. Absorption and noise in x-ray phosphors. J Appl Phys 1973;44:4199–4203.
2. Yaffe MJ, Rowlands JA. X-ray detectors for digital radiography.
Phys Med Biol 1997;42:1–39.
26 Digital Mammography
4