Investigation of combined kV/MV CBCT imaging with a high-DQE MV detector.

PURPOSE: Combined kV-MV cone-beam tomography (CBCT) imaging has been proposed for two potentially important image-guided radiotherapy applications: (a) scan time reduction (STR) and (b) metal artifact reduction (MAR). However, the feasibility of these techniques has been in question due to the low detective quantum efficiencies (DQEs) of commercially available electronic portal imagers (EPIDs). The goal of the work was to test whether a prototype high DQE MV detector can be used to generate acceptable quality pretreatment CBCT images at acceptable dose levels. METHODS: 6MV and 100 kVp projection data were acquired on a Truebeam system (Varian, Palo Alto, CA). The MV data were acquired using a prototype EPID containing two scintillators (a) a standard copper-gadolinium oxysulfide (Cu-GOS) screen having a zero-frequency DQE (DQE(0)) value of 1.4%, and (b) a prototype-focused cadmium tungstate (CWO) pixelated "strip" with a DQE(0) = 22%. The kV data were acquired using the standard onboard imager (DQE(0) = 70%). The angular spacing of the MV projections was 0.81° and the source output was 0.03 MU/projection while the kV projections were acquired with an angular spacing of 0.4° at 0.3 mAs/projection. Image quality was evaluated using (a) an 18-cm diameter electron density phantom (CIRS, Norfolk, VA) with nine contrast inserts and (b) the resolution section of the 20-cm diameter Catphan phantom (The Phantom Laboratory, Greenwich, NY). For the MAR studies, two opposing CIRS phantom inserts were replaced by steel rods. The reconstruction methods were based on combining MV and kV data into one sinogram. The MAR reconstruction utilized mostly kV raw data with only those rays corrupted by metal requiring replacement with MV data (total absorbed dose = 0.7 cGy). For the STR study, projections from partially overlapping 105°kV and MV acquisitions were combined to create a complete dataset that could have been acquired in 18 sec (absorbed dose = 2.5 cGy). MV-only (4.3 cGy) and kV-only (0.3 cGy) images were also reconstructed. RESULTS: The average signal-to-noise ratio (SNR) of the inserts in the MV-only CWO and GOS CIRS phantom images were 0.62× and 0.12× the SNR of the inserts in kV-only image, respectively. The limiting spatial resolutions in the MV-only GOS, MV-only CWO, and kV-only Catphan images were 3, 6, and 8 lp/cm, respectively. In the combined kV/CWO STR reconstruction, all contrast inserts were visible while only two were detectable in the kV/Cu-GOS image due to high levels of noise (average SNRs of kV/CWO and kV/GOS inserts were 0.97× and 0.18× the SNR of the kV-only inserts, respectively). In the kV-MV MAR reconstructions, streaking artifacts were substantially reduced with all inserts becoming clearly visible in the kV/CWO image while only two were visible in the kV/Cu-GOS image (average SNRs of the kV/CWO and kV/Cu-GOS CIRS with metal inserts were 0.94× and 0.35× the SNRs of the kV-only CIRS without metal inserts). CONCLUSIONS: We have demonstrated that a high-DQE MV detector can be applied to generating high-quality combined kV-MV images for SRT and MAR. Clinically acceptable doses were utilized.

SU-E-I-22: Metal Artifact Correction Using KV and Selective MV Imaging.

PURPOSE: To improve the image quality of radiotherapy planning CTs for patients with metal implants or fillings by completing the missing kV projection data with selectively acquired MV data that does not suffer from photon starvation. Using both imaging systems that are available on current radiotherapy devices, streaking artifacts are avoided and the soft tissue contrast is restored, even in areas where the kV photons do not contribute any information. This enables a better delineation of structures of interest in planning CT images for patients with metal objects. METHODS: An algorithm for combining kV and MV projection data from the two on-board imagers of a radiotherapy device is presented in this work. It only requires selective MV imaging with the high energy X-rays being collimated onto the metal implants, ensuring that the patient dose does not increase significantly. The algorithm can cope with non-identical geometries of the two imagers and is based on stitching together kV and MV sinograms by estimating a ratio between them. A numerical head phantom with two dental fillings and two soft tissue patterns was used to quantitatively evaluate the proposed hybrid reconstruction algorithm. A structural similarity index (SSIM) with respect to the ground truth data was computed for two ROIs. Realistic, polychromatic spectra were used for both imagers with 120 keV(p) and 6 MeV(p). The patient dose was limited to about 6 cGy for both acquisitions combined. RESULTS: The reconstruction results yield visually as well as objectively better results (SSIM=74.8%) than a simple sinogram interpolation of the kV data (SSIM=69.7%) or a reconstruction from the original data (SSIM=17.9%). CONCLUSIONS: We have successfully implemented a new reconstruction method for hybrid kV-MV cone beam CT reconstruction that enables a better planning of radiotherapy treatments for patients with metal implants without compromising their safety. This work was funded by NIH grant 1R01CA138426-01A1.

SU-E-I-80: Optimizing Scanning-Beam Digital X-Ray Tomosynthesis of the Lungs.

PURPOSE: We propose to optimize the geometry of the Scanning-Beam Digital Tomography system (SBDX) for application to lung tumor biopsies, thereby providing real-time 3D tomographic reconstructions for target verification. The unique geometry of the system requires trade-offs between patient dose, imaging field of view and tomosynthesis angle. METHODS: We used PCXMC, a Monte Carlo simulation software package, to determine the dose to organs of interest as well as the Average body dose and Effective Dose (both ICRP 60 and 103) for source to detector distances (SDDs) between 90cm and 150cm. To facilitate modeling our system, a modified version of PCXMC was created. We also used matlab to evaluate the possible tomosynthetic angles that Result across the field of view for the same SDDs. RESULTS: To maximize the tomosynthesis angle while leaving space for the patient, an SDD of between 90cm and 110cm is appropriate. At SDD 100cm, patient centered at 40 cm from the detector, operated in fluoro mode, the SBDX system delivers ∼0.38x the dose of a normal mobile fluoroscopy system operating at 30 fps. Because of the inverse geometry of the system, the dose to the patient goes up as the patient gets closer to the detector. Tomosynthetic angles up to 15 degrees over a 5-cm field-of-view can be achieved for this geometry. The patient must be placed within 45cm of the detector in order to achieve the benefits from reduced SDD and increased tomosynthetic angle. CONCLUSIONS: The dose-rate for our optimized geometry is acceptable, although higher dose rates for improved nodule visualization may be required. Additional dose optimization steps include modifying the scanning beam pattern to optimize for tomosynthetic image acquisition. Overall dose during the biopsy procedure will likely decrease since nodule targeting will be improved and the overall number of biopsies required will be reduced. This work has received funding from NIH grant R21 HL098683, as well as from the Lucas Foundation.

First use of a truly-hybrid X-ray/MR imaging system for guidance of brain biopsy.

The use of a new hybrid imaging system for guidance of a brain biopsy is described. The system combines the strengths of MRI (soft-tissue contrast, arbitrary plane selection) with those of x-ray fluoroscopy (high-resolution real-time projection images, clear portrayal of bony structures) and allows switching between the imaging modalities without moving the patient. The biopsy was carried out using x-ray guidance for direction of the needle through the foramen ovale and MR guidance to target the soft-tissue lesion. Appropriate samples were acquired. The system could be particularly effective for guidance of those cases where motion, swelling, resection and other intra-operative anatomical changes cannot be accounted for using traditional stereotactic-based imaging approaches.

Truly hybrid interventional MR/X-ray system: investigation of in vivo applications.

RATIONALE AND OBJECTIVES: The purpose of this study was to provide in vivo demonstrations of the functionality of a truly hybrid interventional x-ray/magnetic resonance (MR) system. MATERIALS AND METHODS: A digital flat-panel x-ray system (1,024(2) array of 200 microm pixels, 30 frames per second) was integrated into an interventional 0.5-T magnet. The hybrid system is capable of MR and x-ray imaging of the same field of view without patient movement. Two intravascular procedures were performed in a 22-kg porcine model: placement of a transjugular intrahepatic portosystemic shunt (TIPS) (x-ray-guided catheterization of the hepatic vein, MR fluoroscopy-guided portal puncture, and x-ray-guided stent placement) and mock chemoembolization (x-ray-guided subselective catheterization of a renal artery branch and MR evaluation of perfused volume). RESULTS: The resolution and frame rate of the x-ray fluoroscopy images were sufficient to visualize and place devices, including nitinol guidewires (0.016-0.035-inch diameter) and stents and a 2.3-F catheter. Fifth-order branches of the renal artery could be seen. The quality of both real-time (3.5 frames per second) and standard MR images was not affected by the x-ray system. During MR-guided TIPS placement, the trocar and the portal vein could be easily visualized, allowing successful puncture from hepatic to portal vein. CONCLUSION: Switching back and forth between x-ray and MR imaging modalities without requiring movement of the patient was demonstrated. The integrated nature of the system could be especially beneficial when x-ray and MR image guidance are used iteratively.

In vitro micronucleus assay with Chinese hamster V79 cells - results of a collaborative study with in situ exposure to 26 chemical substances.

A collaborative study with 10 participating laboratories was conducted to evaluate a test protocol for the performance of the in vitro micronucleus (MN) test using the V79 cell line with one treatment and one sampling time only. A total of 26 coded substances were tested in this study for MN-inducing properties. Three substances were tested by all 10 laboratories and 23 substances were tested by three or four laboratories in parallel. Six aneugenic, 7 clastogenic and 6 non-genotoxic chemicals were uniformly recognised as such by all laboratories. Three chemicals were tested uniformly negative by three laboratories although also clastogenic properties have been reported for these substances. Another set of three clastogenic substances showed inconsistent results and one non-clastogenic substance was found to be positive by one out of three laboratories. Within the study, the applicability of the determination of a proliferation index (PI) as an internal cytotoxicity parameter in comparison with the determination of the mitotic index (MI) was also evaluated. Both parameters were found to be useful for the interpretation of the MN test result with regard to the control of cell cycle kinetics and the mode of action for MN induction. The MN test in vitro was found to be easy to perform and its results were mainly in accordance with results from chromosomal aberration tests in vitro.

Prevention of adriamycin-induced mdr1 gene amplification and expression in mouse leukemia cells by simultaneous treatment with the anti-recombinogen bromovinyldeoxyuridine.

The anti-recombinogenic substance (E)-5-(2-bromovinyl)-2'-deoxyuridine (BVDU) was tested for its ability to prevent adriamycin-induced mdr1 gene amplification and expression in mouse leukemia cells in vitro. F4-6 cells that were treated with stepwise enhanced doses of adriamycin acquired resistance against adriamycin. While 20 ng/ml adriamycin showed strong toxic effects in sensitive cells, the same dose was tolerated at the end of the long-term experiment following treatment with stepwise enhanced doses of adriamycin. In parallel experiments, 0.5 or 1 microg/ml BVDU was given together with adriamycin. BVDU prevented the formation of resistance against adriamycin treatment. Using differential PCR, the signal intensity of the mdr1a-specific band appeared markedly increased in adriamycin-resistant cells, while the signal intensities of the adriamycin + BVDU-treated cells resembled the intensity ratio of the untreated control cells. Beyond that, in resistant F4-6 cells increased expression of mdr genes was demonstrated by Northern blot analysis.

Adaptive resistance of Saccharomyces cerevisiae to chronic treatment with genotoxic and nongenotoxic carcinogens.

The exposure of mammalian cells or tumors for weeks or months to low nonlethal doses of cytostatic drugs may induce multidrug resistance, which can be enhanced by a variety of DNA-damaging agents. Multidrug resistance to a variety of drugs has been observed. But in yeast, DNA-damaging agents have not yet been tested. As the appearance of resistance is the result of longterm exposure, we decided to extend the application of test substances to a period of up to 400 days. In such long-term experiments S. cerevisiae MP1 adapted to treatment with low doses of mutagens. Consistent results were obtained for both genotoxic and nongenotoxic carcinogenic substances, which implies that there may be a single pathway for carcinogens with different modes of action.

Use of Primary Rat and Human Hepatocyte Sandwich Cultures for Activation of Indirect Carcinogens: Monitoring of DNA Strand Breaks and Gene Mutations in Co-cultured Cells.

Loss of cytochrome P-450 content is a common feature in conventional culture systems of primary hepatocytes. In contrast to the standard in vitro situation, in vivo each hepatocyte is exposed to an extracellular matrix (space of Disse) at two opposing basolateral surfaces. This in vivo symmetry has been reconstructed in vitro by culturing rat or human hepatocytes within two layers of collagen, thus forming a sandwich configuration. Activation of dimethylbenzanthracene (DMBA) or benzo[a]pyrene (BaP) was studied in rat and human hepatocytes. Genotoxic effects were studied in a three-dimensional co-culture model between sandwich hepatocytes and mammalian cells using the comet assay for detection of DNA strand breaks, and the HPRT test for detection of gene mutations. Sandwich hepatocytes generated active metabolites. The maintenance of metabolic properties in hepatocytes was dependent on extracellular matrix geometry. The number of DMBA- or BaP-induced genotoxic effects tended to be higher than in standard S-9 mix assays. While the ability to activate indirect carcinogens disappears within hours in primary hepatocytes, hepatocyte sandwich cultures enhance their ability to activate indirect carcinogens within 1 wk and retain this activity for up to 2 wk. This is the main advantage of the sandwich method over the more simple and conventional assays. While freshly isolated hepatocytes, regardless of whether in sandwich culture or in conventional assays, are injured by the isolation procedure and possess a corresponding reduced activation ability, hepatocytes in sandwich cultures recover over the course of a few days, and acquire a much higher ability to activate indirect carcinogens. Consequently, the indirect carcinogens BaP and DMBA, which were ineffective (BaP) or exhibited only weak effects (DMBA) at a concentration of 160nmol/ml in 1-2-day-old hepatocytes, were clearly effective (BaP) or showed about a threefold increase in genotoxicity (DMBA) in 8-day-old hepatocytes in sandwich co-culture. In contrast to the experiments with S-9 mix, which is toxic to mammalian cells and does not allow treatment times of more than 2-3hr, cells in co-culture with human or rat hepatocytes can be treated for at least 24hr. The use of sandwich cultures has not yet been described for genotoxicity studies. The results of the present study may perhaps facilitate the acceptance of this method as a co-culture model for the field of genetic toxicology. Use of hepatocytes alone for genotoxicity studies cannot be recommended for difficulties in isolating intact cells from the sandwich cultures. The use of human hepatocytes in sandwich co-culture should enable a more relevant evaluation of potential human genotoxicity with specific chemicals and should put the extrapolation of genetic toxicology data from animal species to humans on a more scientific basis. Beyond that, experiments with animals in vivo could be avoided.

Reinvestigation of in vivo genotoxicity studies in man. I. No induction of DNA strand breaks in peripheral lymphocytes after metronidazole therapy.

Although a rodent carcinogen, metronidazole is widely used in humans for the treatment of infections with anaerobic organisms. Metronidazole is mutagenic for microorganisms, but has a mainly negative data base for mammals and humans. Therefore, metronidazole is generally considered as a non-genotoxic carcinogen. Only the results of two human in vivo studies would allow the classification of metronidazole as genotoxic carcinogen: (1) the induction of DNA strand breaks; and (2) the induction of chromosome aberrations in peripheral lymphocytes after metronidazole therapy. Because the classification of metronidazole as genotoxic carcinogen would imply enormous consequences with respect to its application, both studies were reinvestigated very thoroughly. The present report describes the reinvestigation of the induction of DNA strand breaks after metronidazole therapy. Each two probes of lymphocytes of metronidazole-treated patients (3 x 500 to 3 x 750 mg/day for 5-8 days) were examined separately for the appearance of DNA strand breaks before and after treatment. In total, 400 nuclei were examined per patient. Immediately before the first, and 30 min to 2 h after the last application, 2 x 10 ml blood per patient was sampled, transported to the laboratory at 15-20 degrees C to make DNA repair more difficult, and examined within the next 4-7 h for DNA strand breaks. At the same time, the individual metronidazole blood plasma levels were measured. In contrast to the published reports, no induction of DNA strand breaks after metronidazole therapy could be observed in the present study. As the applied doses (15,750 mg vs. 4800 mg) and the plasma level (up to 25 micrograms/ml vs. not measured) of metronidazole were much higher than in the published study, the relevance of the clearly negative result is obvious. As induction of DNA strand breaks is a frequent prerequisite for genotoxicity, metronidazole should be considered as a non-genotoxic carcinogen, and not as a genotoxic carcinogen.

Anti-recombinogenic and convertible co-mutagenic effects of (E)-5-(2-bromovinyl)-2'-deoxyuridine (BVDU) and other 5-substituted pyrimidine nucleoside analogs in S. cerevisiae MP1.

In experiments using yeast, without addition of an external metabolic activation system, (E)-5-(2-bromovinyl)-2'-deoxyuridine (BVDU) was co-mutagenic and showed an insignificant anti-recombinogenic effect in combination with triethylene melamine (TEM). In the presence of activating S9-mix, the anti-recombinogenicity and co-mutagenicity could clearly be seen. At higher concentrations the co-mutagenic effect was converted into anti-mutagenicity. The other three 5-substituted pyrimidine nucleoside analogs were tested only in the presence of activating S9-mix and showed similar effects. As TEM is a direct alkylating agent that is inactivated by liver microsomes, the higher activity in presence of S9-mix can be interpreted as resulting from metabolic activation of the 5-substituted pyrimidine nucleoside analogs. In previous experiments using yeast bacteria, Drosophila or mice, tumor promoters were co-recombinogenic/anti-mutagenic, and co-carcinogens were co-mutagenic/anti-recombinogenic. Thus, there is not only an operational difference between tumor promoters and co-carcinogens but a real difference in respect to their genetic effectiveness. As up to now only co-carcinogens have shown co-mutagenic and anti-recombinogenic effects, it is perhaps possible that, within a certain concentration range, 5-substituted pyrimidine nucleoside analogs may have co-carcinogenic activity in carcinogenicity tests. At higher concentrations the co-carcinogenic effect may be converted into an anti-carcinogenic one.

Induction or suppression of SV40 amplification by genotoxic carcinogens, non-genotoxic carcinogens or tumor promoters.

Carcinogens are generally classified into two groups: genotoxic and non-genotoxic. As the final product of genotoxic and non-genotoxic carcinogens is the same, i.e., a clone of genetically altered cells, it could be possible that non-genotoxic carcinogens may yield genotoxic events as a secondary result of cell toxicity having led to mitogenesis/cellular proliferation, or that genetic alterations are induced that are normally neglected in genotoxicity tests. A genetic effect with possible relevance for the ultimate mechanism of carcinogenicity is the amplification of oncogenes. In the present experiments, amplification of SV40-virus DNA in transformed CO631-cells has been measured. In this system, two genotoxic carcinogens (positive control), two non-carcinogens (negative control), and eight non-genotoxic carcinogens have been tested. With the exception of testosterone and BVDU, all substances were tested when given alone. Testosterone and BVDU were tested in combination with either MTX or AMP in the presence or absence of S9-mix. The genotoxic carcinogens TEM and 4-NQO as well as the tumor promoters TPA, coumarin, mezerein and chrysarobin were able to induce SV40 amplification when given alone. The same was for acetamide and thioacetamide. The non-carcinogens acetone and dimethylformamide were ineffective. In case of testosterone, a co-amplification effect was seen in the absence of S9-mix, and an anti-amplification effect in presence of S9-mix. The anti-recombinogen BVDU reduced the SV40 amplification effect of AMP in experiments without addition of S9-mix. In the presence of activating S9-mix, this substance was effective at lower concentrations than without S9-mix, and showed an enhanced anti-amplification effect. Similar effects of testosterone and BVDU had been observed before in respect to stimulation or suppression of recombination. These results can be used as supportive evidence for the assumption that gene amplification is mechanistically related to recombination.

Co-recombinogenic and anti-mutagenic effects of diethylhexylphthalate, inactiveness of pentachlorophenol in the spot test with mice.

In in vivo experiments using the spot test with mice, diethylhexylphthalate (DEHP) was co-recombinogenic and anti-mutagenic. In two independent experiments, DEHP was able to increase in combination with ethylnitrosourea (ENU) the frequency of animals with spots of genetic relevance from about 12% (ENU alone) to about 15% (ENU+DEHP). This enhancement can be exclusively attributed to an enhancement of recombination. In historical as well as in current experiments, with astonishing accuracy, about 8% of all spots induced with ENU alone are black, near-white or twin spots, i.e., presumed products of reciprocal recombination. Under the influence of DEHP, about 20% of all colored spots were black, near-white or twin spots. These co-recombinogenic effects are very strong, considering that the low frequency of twin spots with 0.3% (historical ENU control) or 0.6% (current ENU control) after ENU treatment alone has been enhanced up to a frequency of 3%. While ENU alone induces about 14% light brown colored spots, depending exclusively on gene mutations, under the influence of DEHP the frequency was only 7%, i.e., DEHP was anti-mutagenic. Pentachlorophenol was ineffective in enhancing or reducing recombination or mutations. There was no difference between ENU and ENU+PCP treatment. Since a number of "non-genotoxic' carcinogens with tumor-promoting activities like D-limonene, 12-O-tetradecanoyl-phorbol-13-acetate (TPA) and 2,3,7,8-tetrachloro-dibenzo-p-dioxin (TCDD) have shown similar effects in the spot test, it is suggested that DEHP may have tumor-promoting activity in carcinogenicity tests.

X-ray imaging with amorphous selenium: optimal spectra for digital mammography.

The optimum x-ray spectra for acquisition of digital mammographic images using an amorphous selenium (a-Se) photoconductor are investigated. The recorded images consist of latent charge distributions on the surface of an a-Se plate, which are then read out using two methods, laser discharge, or flat panel recharge. The investigation is based on a model of the breast previously developed for a phosphor-based digital readout system, and has been extended to include the effects specific to the use of photoconductors. The effects of plate thickness, x-ray scatter, readout noise, dose, and the kind of breast tissue on the nature of the optimum spectrum are explored for the two readout methods. The results indicate that use of a kilovoltage setting in the current mammographic range, and a molybdenum target spectrum is appropriate for digital readout of a-Se detectors. This conclusion contrasts with the appreciably higher kilovoltages traditionally used with the xerographic (toner) readout of latent charge images on a-Se.

Anti-mutagenic agents are also co-recombinogenic and can be converted into co-mutagens.

In experiments using yeast without an external metabolic activation system, the hormones testosterone, beta-estradiol, and diethylstilbestrol were anti-mutagenic and co-recombinogenic. In the presence of Aroclor-induced rat liver S-9 plus cofactors (S9-mix) the same substances acted as co-mutagens and anti-recombinogens. Since an external metabolic activation system was able to convert anti-mutagens into co-mutagens, we studied whether the different metabolism of yeast and mice would lead to different effects. In whole-animal experiments using the mouse spot test, anti-mutagenic effects of vanillin (Imanishi et al., 1990), phenobarbital (Shibuya and Murota, 1984), and tannic acid (Sasaki et al., 1990) have been observed. In our present experiments with yeast, tannic acid and phenobarbital showed anti-mutagenic effects also, whereas vanillin acted as a co-mutagen. Thus, as in the case of the hormones, tannic acid and phenobarbital were anti-mutagenic and co-recombinogenic whereas vanillin was co-mutagenic as well as co-recombinogenic.

X-ray imaging with amorphous selenium: detective quantum efficiency of photoconductive receptors for digital mammography.

Factors affecting the zero spatial frequency detective quantum efficiency of photoconductor-based x-ray detectors operating in the mammographic energy range are modeled for monoenergetic incident x rays. The problem is separated into two sections: the calculation of the x-ray absorption and the Swank factor. X-ray absorption in this energy range, for most practical photoconductors, is dominated by the photoelectric effect. The Swank factor has four components: fluorescence escape, stochastic variations in gain, variations of gain due to incomplete coupling of charge from the photoconductive layer to the detector electrode, and the nonlinear discharge arising from the field-dependent x-ray gain, an effect that is unique to photoconductors. Calculations are performed for selenium, which is currently the most technologically advanced photoconductor available for digital x-ray imaging. For thicknesses of selenium exceeding 50 microns and for energies between 12 and 50 keV, the detective quantum efficiency of this photoconductor is found to exceed that of a conventional Gd2O2S-based mammographic phosphor screen.

A model for optimization of spectral shape in digital mammography.

X-ray mammography is the most sensitive imaging modality available for the detection of breast cancer. The highest performance can only be achieved, however, if the complete imaging system is optimized. The development of digital mammography offers an opportunity to obtain improved sensitivity in mammography. In such systems, the decoupling of the recording and display processes allows each component of the imaging system to be optimized separately. In this paper we describe a method for optimizing the recording process for digital mammographic techniques. Our method uses an energy transport model of the propagation of signal and noise through the imaging system. The computations make use of experimentally determined data wherever possible so that the number of assumptions in the model can be minimized. The model predicts the signal-to-noise ratio for a constant dose to the breast, and therefore allows comparison and optimization both for different x-ray spectra and for different imaging tasks. The major energy-dependent components of the model have been verified, and good agreement is demonstrated between predictions by the model of both contrast and SNR and experimentally measured values. Calculations for a particular imaging task, detection of a 200-microns cubic calcification in a 6-cm, 50% adipose-50% glandular breast, illustrate application of the model for optimization of spectral shape.

Optimization of spectral shape in digital mammography: dependence on anode material, breast thickness, and lesion type.

It has been proposed that breast cancer detection can be improved through the use of digital mammography. It is hypothesized that the choice of proper shape of the x-ray spectrum incident upon the breast can yield an improved image signal-to-noise ratio (SNR) for a given dose. To test this hypothesis, an energy transport model incorporating measured breast tissue attenuation coefficients and published exposure-to-dose conversion values was developed to describe the image acquisition process. The choice of applied kilovoltage and filter for Mo and W target x-ray sources has been optimized with respect to SNR and absorbed dose for detectors based on a Gd2O2S scintillating screen under the conditions of perfect coupling of light between the screen and a solid state photodetector. For the W spectra, the optimum filter-kVp combinations could provide 41%, 13%, and 42% improvements in SNR for 2-cm, 6-cm and 8-cm breasts, respectively, over the conventional Mo filtration, for a practical imaging time of 1.0 s. W and Mo spectra produce similar SNR values for a given filter thickness except for the 4-cm breast. Given the limitations of current technology, however, the W spectra produce the optimum SNRs in a shorter imaging time for breast thicknesses greater than and less than 4 cm. The maximum SNR for imaging both infiltrating ductal carcinoma and calcifications is provided by the same filter-kVp combination, allowing optimization based on breast thickness and composition only. The model can now be used to compare and improve upon novel detector designs.

Performance of glass fiber antiscatter devices at mammographic energies.

Using fiber optic manufacturing techniques, it is possible to produce a radiographic grid that discriminates against scattered radiation in two dimensions. Such grids consist of septa composed of glass with a high lead content; the interspace material is air, so that approximately 80% of the grid area is open. In this way, effective high ratio grids can be produced with relatively low Bucky factors. The performance of samples of such grid material is characterized in terms of both scatter rejection and dose efficiency for application in digital mammography in both slot-beam and area-beam geometry. For area beams, five- to tenfold improved scatter rejection relative to conventional grids was observed. In slot configurations, such grids could provide improved SNR/dose performance and more effective utilization of the heat loading capability of the x-ray source.

Dynamic range requirements in digital mammography.

The dynamic range and the number of gray levels, gamma s, required for digital mammography has been evaluated using an energy transport model. The effects of molybdenum (Mo) and tungsten (W) target spectra and the energy-dependent attenuation by elemental filters, breast tissue, and a phosphor screen were included in the model. For detectors with ideal optical coupling and no inherent detector noise, 3,100 gray levels are discernable (requiring 12 bits per pixel), assuming a 40 kVp, W target spectrum (1.0 mm A1 filtration), a mean glandular dose to a 5 cm thick breast of 0.6 mGy, and an ideal observer with a 5 mm diam viewing aperture. The effects of inherent detector noise and realistic coupling efficiency on gamma s were also examined. For the 40 kVp, W spectrum, a detector with total coupling efficiency of 16 electrons (e-) per x-ray interaction and a dynamic range of 3000 (maximum carrier signal of 1.93 x 10(5) e-/pixel and inherent detector noise of 64 e- pixel) would decrease the number of gray levels that could be resolved by only 2% compared to a detector with ideal coupling and no inherent noise. A detector with a total coupling efficiency of 2.0 electrons per x-ray interaction and a dynamic range of 240 (maximum carrier signal 2.41 x 10(4) e-/pixel and inherent detector noise of 100 e-/pixel) would reduce the number of gray levels by 26% for the 40 kVp spectrum. On the basis of dynamic range, W spectra are preferable for digital mammography, since Mo spectra yielding the same signal-to-noise ratio require a detector with dynamic range twice as large, and with a 30% greater saturation signal. When no scatter rejection method is used, scattered radiation over a 254 cm2 imaging field reduces the number of discernable gray levels by 23% for a 5 cm thick breast and 34% for an 8 cm thick breast.