Mammography is a specialized x-ray imaging technique

Transcription

1 ORIGINAL ARTICLE Rüdiger Lawaczeck, PhD,* Vladimir Arkadiev, PhD, Felix Diekmann, MD, and Michael Krumrey, PhD Received April 29, 2004 and accepted for publication, after revision, September 12, From *Schering AG, Forschungslaboratorien, Berlin, Germany; Institut für Gerätebau GmbH, Berlin Adlershof, Germany; Universitätsklinikum Charité, Institut für Radiologie, Berlin, Germany; and Physikalisch- Technische Bundesanstalt, Berlin, Germany. The project was supported by the City of Berlin Program for the Development of Industrial Technology (FiTE) in Berlin and by the European Fonds for Regional Development (EFRE) under the project number Correspondence: Rüdiger Lawaczeck, PhD, Schering AG, Research Laboratories Berlin, Müllerstr. 178, Berlin, Germany. ruediger. Copyright 2004 by Lippincott Williams & Wilkins ISSN: /05/ Rationale and Background: The emission spectrum of an x-ray tube is determined by the anode and filter materials as well as by the high voltage being used. For mammography, typical anode materials are molybdenum (Mo), rhodium (Rh), and tungsten (W); molybdenum, rhodium, and aluminum are favored for filters. Mammography is a soft tissue imaging modality demanding a high spatial resolution as well as a high detector sensitivity. Low-energy photons are only absorbed in tissue and have no contribution to the image; nevertheless, they increase the dose. High-energy photons mostly penetrate soft tissue and generate a background noise as a result of strong scattering that deteriorates the image quality. For mammography, the optimal energy window is in a range from 17 and 25 kev. From a theoretical perspective, one would favor monoenergetic x-rays (eg, the Mo-emission line at 17.5 kev). This article presents the realization of imaging with monochromatic x-rays using a diagnostic mammography unit. Methods: Basically, a monochromatic module was added to a conventional mammographic system. The monochromatic module can be mounted at the end of the x-ray tube and it consists of a curved HOPG (highly oriented pyrolytic graphite) crystal and a slit collimator. For image generation, the object is moved through the fan-shaped monochromatic radiation field. In addition to the conventional polychromatic 2-dimensional case, the polychromatic irradiation was also able to be performed under similar conditions. For image acquisition, image plates or a linear array detector were used. Exposure doses were measured for both poly- and monochromatic radiation. The initial evaluation of the system performance was carried out by imaging a contrast-detail phantom and biologic specimens. Results: The monochromatic x-ray beam has a size of approximately 35 mm 200 mm in the object plane. The photon flux of the monochromatic x-rays is considerably lower than the photon flux of the polychromatic x-rays but adequate for initial studies of phantoms, biologic tissue, or small animals. The comparison of the results obtained with the monochromatic and polychromatic imaging modalities reveal a conspicuous increase of image contrast in the monochromatic case. Conclusion: The results suggest that the experimental setup for monochromatic excitation shows clear potentials for improvements of the image in comparison to the conventional polychromatic case. Key Words: monochromatic module, mammography, contrast enhancement, phantoms and biologic specimens (Invest Radiol 2005;40: 33 39) Mammography is a specialized x-ray imaging technique for the detection of tumors of the female breast (for review of the development, see reference 1 ) with demands on a high image quality, in particular in terms of contrast and detail resolution. Mammography is a soft tissue imaging technique. The radiation path is limited to 4 to 5 cm on average. Because, in the absence of K-edges, tissue contrast decreases with increasing x-ray photon energy, the mammography is performed with low-energy x-rays. Mo-anode or bi-anode combinations of Mo and W or Mo and Rh are common with high voltages in the range between 22 and 35 kv; in some cases, 49 kv are available. Mo, Rh, or Al filters additionally reduce the low energy part of the x-ray spectra. Mammography is supported by palpation or ultrasound diagnostics and in critical questions complemented by magnetic resonance mammography subsequent to the administration of Gd-containing contrast media. Currently, flat-panel detectors are replacing the film-foil system. Although the spatial resolution of the film-foil system is superior, flat-panel detectors open the way to digital mammography with the complete repertoire of software image processing. 2 6 Conventional medical x-ray tubes emit a broad Bremsspectrum with anode-characteristic emission lines superimposed, eg, Mo K at 17.5 kev or Rh K at 20.2 kev (see, for example, reference 7 ). In soft tissue imaging, the low-energy x-rays do not contribute to contrast but increase the radiation dose because all photons are absorbed. On the other side, the high-energy photons are almost exclusively scattered and Investigative Radiology Volume 40, Number 1, January

2 Lawaczeck et al Investigative Radiology Volume 40, Number 1, January 2005 thus decrease image quality. Optimal energies in mammography are between 17 and 25 kev. On the excitation side, the absorption filtering 8 reduces the low-energy contribution but filtering can only partially narrow the spectrum to the optimal range. A clear choice of the optimal x-ray energy is only possible with the help of monochromators. For that reason, a crystal-based monochromator was designed and adapted to a conventional mammography unit. Optimal signal intensities were reached by tuning the monochromator to the Mo emission line of the bifocal Mo-anode tube used. The main aim of the studies described was to modify a standard mammography setup in such a way that on the excitation side, monochromatic x-rays can be used with all other components not being changed so that clinically oriented studies can be performed. The intention was not to use highly monochromatized x-ray beams from synchrotron radiation, but to find a compromise between monochromaticity and photon flux with conventional x-ray tubes. Because a modular platform was favored, the monochromator can easily be removed or reinstalled so that comparative studies, ie, monochromatic versus polychromatic excitation, are straightforward. The monochromator module produces a stripe of monochromatic radiation so that either the object must be moved through the beam or the beam must be scanned over the object to acquire the image of the whole object. In the latter variant, the system resembles the slot-scan system in which the x-ray beam moves over the compressed breast in synchrony with the detector under the breast. 6,9 The present approach could considerably profit from the use of a curved crystal as a dispersing element of the monochromator module in combination with a linear array detector with high spatial resolution and dynamic range described in detail by Tesic et al. 9 Initial results focusing on the CD-MAM phantom have been recently reported by Diekmann et al. 10 Gambaccini and coworkers 11 also pursued monochromatic x-rays for mammography using HOPG crystals in a different arrangement. Carroll and colleagues elaborated on pulsed, tunable, monochromatic x-ray beams generated by Compton backscattering of laser radiation at a high-energy electron beam. 12,13 These narrow-beam monochromatic x-rays could be developed into a compact source for future studies in mammography. 12 MATERIALS AND METHODS Overall System The system used consists of a Siemens Mammomat 300 (x-ray tube with a bifocal molybdenum anode, generator, console, and mainframe) with 2 additional modules: an x-ray monochromator, which could be installed at the exit of the x-ray tube, and a linear array detector, which was used instead of the film/foil system. Figure 1 gives an overview over the complete system. For all measurements, the high voltage was set to 35 kv. One hundred forty mas or 400 mas were chosen for the small (0.15 mm) or large (0.3 mm) focus with the respective currents of 20 and 100 ma so that the operation time is 7 seconds and 4 seconds for the small and large focus, respectively. Monochromator Module The monochromator is an experimental module, which can be optionally installed at the exit of the x-ray tube. The FIGURE 1. Monochromator module mounted at the exit of the x-ray tube (left) and overview of the whole system with Mammomat main frame and monochromator module. The system is tilted 12 so that the monochromatic x-ray beam falls vertically onto the linear array detector. The object, 2 plexiglass plates with a Pb line grating in the present case, is mounted on a carriage, which is moved at constant velocity over the detector system through the radiation field. The conventional film cassette holder is removed out of the radiation field and just visible at the left lower corner Lippincott Williams & Wilkins

3 Investigative Radiology Volume 40, Number 1, January 2005 x-rays leaving the x-ray tube fall under the Bragg angle on a curved HOPG crystal. The collimator slit in the focal plane of the crystal acts as virtual x-ray source. The geometry was optimized for the K emission line of the molybdenum anode (17.5 kev). The 2d spacing of the (002) planes in HOPG (highly oriented pyrolytic graphite) crystals is Å, so that the Bragg angle for the energy of 17.5 kev amounts to 6.1. The energy resolution is better than 5%, so that only the Mo-K emission line passes the monochromator but not the Mo-K line. The radiation path is outlined in Figure 2; the physical parameters are summarized in Table 1. Detector The linear array detector (TDI CCD Th9570; Thales Electron Devices SA, Moirans, France) is used in the time-delay integration (TDI) modus. 9 Minimal pixel size is 27 m 27 m with 12-bit sensitivity. However, in most cases, the binning mode with 4 neighboring pixels being combined was preferred as a result of the higher sensitivity. The detector has 400 rows and 8184 columns corresponding to a size of 10.8 mm mm. Some images were obtained also with AGFA image plates with the same 12-bit sensitivity but with a considerably lower local resolution under 100 m. Image Generation and Acquisition The linear array detector is used in the TDI mode. 9 In that mode, the velocity of the charge transfer from 1 row to the neighbor row in the detector unit and the velocity of the object movement have to be synchronized strictly. In the present case, the velocity of the object movement is determined by a softwarecontrolled stepper motor. The digital signal from the linear array detector is connected to a dedicated computer program, which generates and displays the image as a 2-dimensional intensity graph. Two corrections of the image are necessary to handle the different sensitivities of the detector and inhomogeneities of the x-ray beam. First, the offset signal intensity I 0 of the detector in the absence of radiation is measured and stored. Second, a reference signal intensity I ref is measured, which represents a blank image without object but contains all the inhomogeneities FIGURE 2. Radiation path through the monochromator module. x c, y c, and x refer to the coordinate systems in the collimator and object planes respectively. TABLE 1. Module Physical Parameters of the Monochromator Crystal: HOPG (Optigraph Ltd., Moscow, Russia) Mosaicity: 0.4 Crystal length: 50 mm Crystal width: 20 mm Curvature radius: 480 mm Distance tube-focus HOPG crystal: 50 mm Distance crystal collimator: 50 mm Collimator slit: 0.3 mm (variable) Distance collimator object plane: 550 mm (variable) of the monochromatic beam. This image is also stored. Third, the raw image of the object I obj is measured and stored. Finally, the corrected image of the object I cor is calculated from the 3 images according to I cor (I obj I 0 )/(I ref I 0 ). On that basis, different sensitivities of individual detector pixels and inhomogeneities of the radiation field can be compensated and corrected. The images can be further processed by the specially developed software. Alternatively, the image data can be stored in dicom format and processed by common image routines. Characterization of the Radiation Monochromaticity and homogeneity of the monochromatic beam in the object plane were measured by a Si(Li) detector (Röntec AG, Berlin, Germany), by a CdZnTe detector (AMPTEK, Bedford, MA), as well as by a system based on phosphor screen, tandem optics, and a CCD camera. For absolute measurements of the spectral irradiance, an energydispersive Si(Li) detector was calibrated with synchrotron radiation using the electron storage ring BESSY II as primary radiometric source standard. 14 The energy calibration and the response function of the detector were obtained at an x-ray facility equipped with a high-resolution monochromator. The transmission of different filters used for count rate reduction in the polychromatic beam was determined with synchrotron radiation at a wavelength shifter beamline. 15 The measured spectral irradiance was used to calculate the absorbed dose. For additional measurements of the surface dose, a clinical instrument (Unidos PTW, Freiburg, Germany) was available. Imaging Objects Objects were the CD-MAM phantom, combinations of plastic plates, and biologic tissue. The CD-MAM phantom (Med. Dept. University of Nijmegen, Netherlands) consists of a plexiglass plate (0.5-cm thick) with 205 embedded circular gold dots ranging from 1.25 m to 0.06 m in thickness and 2004 Lippincott Williams & Wilkins 35

4 Lawaczeck et al Investigative Radiology Volume 40, Number 1, January 2005 from 3.2 mm to 0.1 mm in size. The phantom was complemented with each 2-cm PMMA on both sides. In addition to the CD-MAM phantom, plexiglass plates with 1-cm plastic cuvettes containing solutions of the contrasting elements Br, Y, Zr, I, and Gd (ordered according to increasing atomic number) were studied. As a biologic specimen, sheep breast tissue with inlets containing agar plates additionally filled with I and Yb was studied. In vivo studies were performed with rats in consent with German regulations. General Procedure For studies with monochromatic radiation, the frame with the x-ray tube is tilted 12 so that the monochromatic beam vertically strikes the array detector. At the same time, a polychromatic direct beam (not reflected by the HOPG crystal) transverses the collimator and is stopped by a lead foil. If the frame is moved back to the 0 position, this direct beam falls vertically on the detector, allowing generation of polychromatic images with the slot-scan geometry comparable to the monochromatic case. For comparison, the monochromator module was removed and images were generated also with the standard radiation field. RESULTS Figure 3 shows the result of the measurement of the radiation homogeneity. In the object plane at a distance of 550 mm from the collimator, the x-ray beam has a size of approximately 35 mm 200 mm. The homogeneous size of the illumination is approximately 10 mm in the x-direction; in y-direction, the homogeneity corresponds to the 200 mm illuminated but the intensity falls off at the edges. The homogeneous region of the monochromatic beam practically covers the linear array detector. For the optimally positioned monochromator module, only the intensity but not the monochromaticity is space-dependent. The spectral irradiance for the polychromatic and the monochromatic beam is shown in Figure 4. The measurements were performed with the calibrated Si(Li) detector. The polychromatic spectrum is dominated by the Mo K and K lines at 17.5 kev and 19.6 kev in addition to the broad Bremsstrahlung contribution. With the monochromator, the Mo K radiation can be isolated. From the measured spectral irradiance, the spectral entrance dose rate has been calculated using literature data for the mass-energy absorption coefficients. 16 Taking into account a model in which the compressed breast is composed of 50% glandular and 50% adipose tissue FIGURE 3. Homogeneity of the monochromatic illumination (part grey corresponds to highest, black to lowest intensity). 36 FIGURE 4. Spectra of the polychromatic and the monochromatic radiation measured with a calibrated Si(Li) detector. Only the Mo K radiation passes the monochromator. and is surrounded by a 0.5-cm thick adipose layer, 17 the mean glandular dose (MGD) rate has been calculated from the entrance dose rate. The obtained MGD rate for monochromatic radiation is lower by approximately a factor of 12 compared with the polychromatic radiation. Thus, for the same MGD, the exposure time in the monochromatic mode can be increased 12 times to reach the polychromatic case. Spectra of the polychromatic x-rays before and after passing 4-cm plexiglass are shown in Figure 5. The radiation is hardened because the low-energy photons are absorbed and only increase the dose but do not contribute to the image. In contrary to the polychromatic case, monochromatic radiation is only attenuated by the plexiglass without change of the spectrum. The attenuation of solutions containing the contrasting elements Br, Gd, I, Y, and Zr are summarized in Figure 6 in which the absorption with respect to water is shown. Gray values, or pixel intensities I(x), were read from image plates and ln(i(water)/i(x)) values were calculated subsequently. The absorption decreases in the order: Gd Y Br Zr I. Br and Y have the K-edges at 13.5 and 17.1 kev, respectively, just below 17.5 kev. Zr has the K-edge at 18 kev just above 17.5 kev and the K-edges of I and Gd at 33.2 and 50.2 kev are far above the 17.5 kev line. The absorption values for polychromatic and monochromatic x-rays are in the same range. A considerable difference is only observed for the Zr-solution in which for monochromatic x-rays at 17.5 kev, the attenuation coefficient is low. Because the polychromatic spectrum overlaps with the K-edge, it leads to a higher absorption. Images of biologic tissue, in which Figure 7 is an example, reveal that in both cases, small calcifications ( 0.5 mm) can be easily visualized. The visual inspection of the agar inlets reveals that the contrast with respect to the breast tissue is more pronounced for the monochromatic x-rays than for the polychromatic x-rays. This trend is also seen in Figure 8 in which the same part of the CD-MAM phantom is 2004 Lippincott Williams & Wilkins

5 Investigative Radiology Volume 40, Number 1, January 2005 FIGURE 7. Images of sheep breast tissue compressed to 30-mm thickness between 2 3-mm plexiglass plates. (A) monochromatic x-rays, (B) polychromatic x-rays. AGFA image plates. The agar inlets have a diameter of 17 mm and contain, from left to right: control water, Yb 0.1 mol/l, Yb 0.02 mol/l, I 0.1 mol/l, and I 0.02 mol/l. FIGURE 5. Spectra of the polychromatic x-ray beam from a Mo anode at 35 kev with Mo filters before (top) and after (bottom) passing 4-cm plexiglass. All photons with energy below 15 kev are absorbed in the plexiglass and would not contribute to the image. FIGURE 8. CD-MAM phantom (top) and identical details (bottom) obtained with polychromatic x-rays (left) and monochromatic x-rays (right). Surface doses are approximately 18 mgy for the polychromatic case and 9 mgy for the monochromatic case, respectively (Diekmann et al 10 ). FIGURE 6. X-ray absorption ln(i(water)/i(x)) /2 for 0.2 molar aqueous solutions of Br, Gd, I, Y, Zr (KBr, Gadovist, Ultravist, YCl 3, and ZrOCl 2 ) and water for monochromatic and polychromatic x-rays, respectively. Ten-millimeter plastic cuvettes; gray values were read from AGFA image plates. The factor 0.5 results from the square root compression of the data. reproduced for the polychromatic and monochromatic x-rays. Although as a result of the lower photon flux in the monochromatic case, the signal/noise ratio is considerably lower for monochromatic x-rays, the contrast is higher. This is quantitatively shown in Figure 9 in which, from images similar to those in Figure 8, the contrast C is calculated from the intensities I according to C (I background I golddot )/I background When the contrast C is plotted as function of the thickness of the gold dots, 2 straight lines are obtained for the 2 cases 2004 Lippincott Williams & Wilkins 37

6 Lawaczeck et al Investigative Radiology Volume 40, Number 1, January 2005 FIGURE 9. Calculated contrast values as function of gold thickness from CD-MAM images obtained with polychromatic and monochromatic x-rays. Slopes of linear regression lines are and (counts/mm) for the polychromatic and monochromatic case, respectively. Contrast intensity(background) intensity(gold) /intensity(background). regarded as seen in Figure 9. The linear dependence of the contrast on the thickness of the gold dots holds as long as the absorption of gold dots over the background is small. The slopes of the straight lines differ by almost a factor of 2 in favor of the monochromatic x-rays. This behavior is almost independent of the radiation dose and holds as long as the signal/noise ratio is not limiting, which can be the case for rather small objects and low x-ray fluxes. DISCUSSION The experimental setup considered in the present studies allows investigations with monochromatic x-rays under conditions close to the mammographic setting without using an expensive and laborious synchrotron radiation source like in other studies The monochromator was designed for adaptation to an available mammographic system and can be easily mounted at the exit of the x-ray tube or removed for comparative studies with conventional polychromatic x-rays. Instead of an array of flat crystals as described earlier, 11 only 1 curved HOPG crystal serves as a dispersive element. A collimator slit in the focal plane of the crystal acts as a virtual x-ray source, which determines together with the detector the spatial resolution of the system. The monochromatized beam leaves the collimator slit, passes through the object, and falls onto a high-resolution digital linear array detector. For image generation, the object must be moved through the beam, or the beam in synchrony with the detector must be scanned over the object. In this respect, the system is comparable to slot-scan systems using the identical detector. 9 Currently, the entrance dose of the monochromatic radiation is below 1 mgy, whereas under identical experimental conditions, the polychromatic entrance dose can be as high as 36 mgy. The high voltage was always set to 35 kv. Because the current is fixed by the manufacturer to 20 ma for the small focus 38 and to 100 ma for the large focus, respectively, long scan times can only be achieved by choosing the maximal mas products (140 or 400 mas) resulting in scan times of 7 seconds or 4 seconds, respectively. This time was sufficient for acquiring large images of standard phantoms in the TDI mode. A high spectral purity of the monochromatized beam was observed with a single peak at 17.5 kev. The beam is 200 mm long and 35 mm wide at a distance of 550 mm from the collimator slit. In the longitudinal direction, the beam profile and the monochromaticity do not dramatically change, although the intensity falls off at the edges. In the perpendicular direction, the beam is monochromatic and homogeneous over the central part of approximately 10 mm. Studies with contrast media at equal molar concentrations reveal for the elements studied a decrease of absorption in the series Gd Y Br Zr I. The same order was observed in previous studies. 7 The series can be explained on the basis of the energy-dependent attenuation coefficients. Except for Zr with a K-edge at 18 kev, the attenuation of the elements is not remarkably different between monochromatic and polychromatic x-rays. By going from monochromatic to polychromatic x-rays, the spectral mean value slightly increases and starts to overlap with the K-edge of Zr leading to a pronounced increase in attenuation for polychromatic radiation. It was shown by measurements with phantom and breast specimens that an excellent spatial resolution can be obtained and a contrast improvement by a factor of 2 is possible with help of monochromatic x-rays (see also reference 22 ). However, at the rather low doses for the monochromatic x-rays, the low signal/noise ratio can become critical so that the detail contrast can fall below that obtained with conventional x-rays. One way to circumvent the low signal/ noise ratio for the monochromatic case is the summation of a number of separately obtained images. Initial findings of this project have been recently reported, the focus being on the contrast detail of the CD-MAM phantom. When almost equivalent radiation doses were used, the monochromatic technique resulted in slightly superior detail contrast in comparison with polychromatic techinques. 10 In the present study, the entrance dose was measured and it was found that the surface doses of monochromatic and polychromatic x-rays can differ by almost a factor of 60. This factor critically depends on the geometric adjustment of the monochromator module. For the monochromatic case, the dose profile over the phantom thickness strictly follows the exponential attenuation law, whereas deviations from the exponential behavior are observed for conventional polychromatic x-rays. Polychromatic x-rays are hardened on the way through the object and thus deposit a higher dose, especially in the surface near layers. Two ways to overcome the low flux problem of the monochromatic x-rays are currently considered: 1) tubes with higher power should be made available from the manufac Lippincott Williams & Wilkins

7 Investigative Radiology Volume 40, Number 1, January 2005 turer side or 2) the object is scanned several times and subsequently the pixel values are summed. For the studies presented here, the latter alternative was tested, although this approach is clinically not practicable. The results presented suggest that monochromatic x-ray mammography is a promising technique with improved contrast as has been already shown in earlier, partly very complex studies of monochromatic radiation in mammography In combination with a high-resolution detector, monochromatic radiation has a potential for providing images with a high contrast and a high spatial resolution at identical doses compared with conventional mammography. Carroll et al 23 found that breast specimens of normal and cancerous tissues can show slightly different attenuation of monochromatic x-rays. Thus, it is not only possible to reduce doses but also for the diagnostic value of x-ray mammography to profit from the use of monochromatic x-rays and gain specificity. In the present case, the object is moved softwarecontrolled through the fan-beam radiation field with the source and detector being fixed in space. For clinical investigations, it seems to be more appropriate to turn the radiation source and detector over the compressed breast similar to the digital slot scan mammographic units. 9 The axis of rotation could then coincide with the virtual focus of the collimator slit. CONCLUSION It has been shown that a conventional mammography instrument can be considerably improved by adding 2 modules: 1) Spatial resolution up to 27 m can be gained if a digital linear array detector is used instead of the film/foil or digital flat-panel detector system. This arrangement is verified in digital slot/scan instruments. 2) Imaging with monochromatic radiation becomes possible if a monochromator is mounted at the exit of the x-ray tube. The heart of the monochromator module described is a curved HOPG crystal focusing the monochromatic x-rays onto the slit collimator, which acts as a virtual x-ray source. Studies with phantoms, contrast media, and biologic specimens reveal that the contrast in the case of monochromatic x-rays is considerably higher as achieved with polychromatic radiation. Currently, the low intensity of the monochromatic radiation field and consequently a low signal/noise ratio are the rate-limiting factors. Increasing the intensity should be the aim of further investigations, which demand the optimization of all the components of the system but, first of all, increasing the power of the x-ray tube. ACKNOWLEDGMENTS In addition to the authors, the following colleagues participated in the MCR (monochromatic röntgen) project: Aniouar Bjeoumikhov, PhD, Johannes Rabe, Peter Roth, PhD, Joachim Tilgner, PhD, Reiner Wedell, PhD, and Norbert Langhoff, PhD, from the Institut für Gerätebau GmbH (IfG) Berlin-Adlershof; Ulrike Linke and Gerhard Ulm, PhD, of the Physikalisch- Technische Bundesanstalt (PTB) Berlin; Kari Richter, MD, and Bernd Hamm, MD, from the Institut für Radiologie, Universitätsklinikum Charité Berlin; Wolfhard Semmler, MD, PhD, from the former Institut für Diagnostikforschung GmbH (IDF) Berlin; and Wolf-Rüdiger Press, Katja Schön, MD, Angelika Tölle, PhD, and Hanns-Joachim Weinmann, PhD, from Schering AG, Research Laboratories Berlin. REFERENCES 1. Sickles EA. Breast imaging: from 1965 to the present. Radiology. 2000;215: Skarpathiotakis M, Yaffe MJ, Bloomquist AK, et al. Development of contrast digital mammography. Med Phys. 2002;29: Lewin JM, Isaacs PK, Vance V, et al. Dual-energy contrast-enhanced digital subtraction mammography: feasibility. Radiology. 2003;229: Haus AG, Yaffe MJ. Screen-film and digital mammography. Image quality and radiation dose considerations. Radiol Clin North Am. 2000; 38: Shah AJ, Wang J, Yamada T, et al. Digital mammography: a review of technical development and clinical applications. Clin Breast Cancer. 2003;4: Pisano ED, Yaffe MJ, Hemminger BM, et al. Current status of full-field digital mammography. Acad Radiol. 2000;7: Lawaczeck R, Diekmann F, Diekmann S, et al. New contrast media designed for X-ray energy subtraction imaging in digital mammography. Invest Radiol. 2003;38: Sandborg M, Carlsson CA, Carlsson GA. Shaping x-ray spectra with filters in x-ray diagnostics. Med Biol Eng Comput. 1994;32: Tesic MM, Fisher Piccaro M, Munier B. Full field digital mammography scanner. Eur J Radiol. 1997;31: Diekmann F, Diekmann S, Richter K, et al. Near monochromatic x-rays for digital slot-scan mammography: initial findings. Eur Radiol. 2004; 14: Gambaccini M, Tuffanelli A, Taibi A, et al. Spatial resolution measurements in quasimonochromatic x-rays with mosaic crystals for mammography application. Med Phys. 2001;28: Carroll FE. Tunable monochromatic x rays: a new paradigm in medicine. AJR Am J Roentgenol. 2002;179: Carroll FE, Mendenhall MH, Traeger RH, et al. Pulsed tunable monochromatic x-ray beams from a compact source: new opportunities. AJR Am J Roentgenol. 2003;181: Thornagel R, Klein R, Ulm G. The electron storage ring BESSY II as a primary source standard from the visible to the x-ray range. Metrologia. 2001;38: Görner W, Hentschel MP, Müller BR, et al: BAMline: the first hard x-ray beamline at BESSY II. Nucl Instr Meth A. 2001; , Selzer SM. Calculation of photon mass energy-transfer and mass energyabsorption coefficients. Radiat Res. 1993;136: Arfelli F, Bonvicini V, Bravin A, et al. Mammography of a phantom and breast tissue with synchrotron radiation and a linear-array silicon detector. Radiology. 1998;208: Burattini E, Cossu E, Di Maggio C, et al. Mammography with synchrotron radiation. Radiology. 1995;195: Johnston RE, Washburn D, Pisano E, et al. Mammographic phantom studies with synchrotron radiation. Radiology. 1996;200: Moeckli R, Verdun FR, Fiedler S, et al. Objective comparison of image quality and dose between conventional and synchrotron radiation mammography. Phys Med Biol. 2000;45: Thomlinson W, Berkvens P, Berruyer G, et al. Research at the European Synchrotron Radiation Facility medical beamline. Cell Mol Biol (Noisyle-grand). 2000;46: MacDonald CA, Gibson WM. Applications and advances in polycapillary optics. X-Ray Spectrom. 2003;32: Carroll FE, Waters JW, Andrews WW, et al. Attenuation of monochromatic x-rays by normal and abnormal breast tissues. Invest Radiol. 1994;29: Lippincott Williams & Wilkins 39