The generation of high quality images made mammography the chosen method to identify early breast cancer. The success of the digital mammography depends upon the production of high quality images with optimized doses that result in an improved cost-benefit relationship to the patients.[1,2] The digital mammography begins to gain space into the Brazilian market and doubts appear among users, in terms of advantages of the use of this technology, as well as risks associated to the radiation doses, when compared to conventional image systems.[3]

Today, most research related to digital mammography presents comparative parameters with conventional methods, due to this last one being the reference system in breast injury detections in early stages.[4]

The detection process of suspicious structures in the image have been a great concern to specialists. Many factors can influence the mammographic image quality and, therefore, the accuracy of the diagnostic.

To keep a good quality standard of the mammographic image, some requirements are established by the Brazilian legislation to verify changes in the image generation system.[3] Quality verification tests from the Quality Assurance Program (QAP) must be capable to assure the detectability of tiny objects with radiation doses lower than or equal to the reference levels acceptable by the QAP. For such tests, several types of mammographic phantoms are used.[5]

In fact, there are many variables to determine the dose that can modify the quality of the image and, among them we can note the thickness and composition of the breast, radiation quality, and scattered radiation intensity. The entrance skin dose (ESD) is useful as a reference to compare the doses in different exposure parameters.

Three hundred images were generated from Alvim Statistical Phanton model 18-209, on 5 CR systems from different manufacturers, using the same equipment model Senographe DMR-GE, on operation modes STD, CNT and DOSE. The CR systems used were: Fuji model FCR Profect CS 50 and FCR XG 5000, Kodak model 875 and Agfa model CR 75-X and CR 85-X.

Half of the images were generated adding 2 cm acrylic over the phantom to simulate a thick or big breast. Therefore, the phantom images were generated to represent breasts with normal thickness (4.5 cm) or higher (6.5 cm).

During the realization of the images, the doses on skin entrance were measured, according to the ACR’s protocol, using ionization chamber Innovision and the simulator ALVIM. The images were presented on the high resolution monitor Barco (5.0 Mpixels and luminance?500 cd/m2) through software developed for training making digital tools available (contrast and brightness adjust, amplification, and inversion of the gray levels). Through this software, it is possible to automatically analyze the reader’s responses and his performance based on the signals detection theory (ROC curve), thru indicators: probability of detectability (Pdet), kappa values (k), true positive results (VP), false positive results (FP), and area beneath the ROC curve for microcalcifications and fibers of different sizes present into the phantom.[6]

The readings were taken by 4 trained professionals on digital image readings in a controlled environment with regard to luminosity (? 20 lux) and noises that could cause missed concentration. During the readings, the professionals were focused on the detection of the microcalcifications and fibers. For such, each finding was classified by the professional in 5 levels of confidence: 100% (the object is definitely there), 75% (the object is likely there), 50% (presence of the object is uncertain), 25% (presence of the object is unlikely) and 0% (there definitely is no object).

The detectability of the microcalcifications and fibers and the degree of concordance of the readings with the true position of simulated structures identified (kappa values) were calculated automatically.

Figure 1 shows the results of the mean detection probability (Pdet) of professionals when detecting microcalcifications (M) and fibers (F), considering all the operational modes (CNT, STD and DOSE) in the different CR, for normal breasts, as well as the ESD.

Figure 2 shows the results of the mean Kappa values (K) of professionals when detecting microcalcifications (M) and fibers (F), considering all the operational modes (CNT, STD and DOSE) in the different CR, for normal breasts, as well as the ESD.

Figure 3 shows the results of the average detection probability (Pdet) of professionals when detecting microcalcifications (M) and fibers (F), considering all the operational modes (CNT, STD and DOSE) in the different CR, for thicknesses breasts as well as ESD.

Figure 4 shows the results of the average Kappa values (K) of professionals when detecting microcalcifications (M) and fibers (F), considering all the operational modes (CNT, STD and DOSE) in the different CR, for thicknesses breasts, as well as the ESD.

The CR systems were installed by manufacturers in the same diagnoses clinic and the images were taken the same week during the selection of digital mammography system. The adjustments parameters of the CR to the mammographic equipment were made by each manufacturer with the objective being to get the best signal-to-noise ratio associated to the lowest radiation dose. Comparing the kappa values for normal and thicker breasts, it is possible to observe a decreasing Kappa value of 12.5% and an increase of 80% for ESD, except the Kodak 975. In this last one, the ESD was similar for different thicknesses, while kappa value decreased of 21,6%. The better reading performance of radiologists was gotten in the FCR Profect and CR85X mammographic digital system with similar ESD. There wasn’t a correlation between the readings performance and ESD for the CR Kodak 975, suggesting that this system probably was not well calibrated.

The kappa values show differences of the readings performance of radiologists between the digital mammographic systems.

The skin entrance dose for CR systems were comparable to screen-film mammography and below the reference levels suggested by the Brazilian QAP (10 mGy).

1. Pisano ED, Yaffe MJ, Digital Mammography. Radiology. 2005;234:353–362.
2. Young KC, Ramsdale ML, Rust A, Cooke J. Effect of automatic kV selection on dose and contrast for mammographic X-ray system. B J Radiol. 1997;70:1036-1042.
3. Payne M, Lawinski PA. Comparison of Four Mammographic Image Quality Test Objects. Br J Radiol. 1992;65:339-341.

4. Krug KB, Stutzer H, Girnus R, et al. Image Quality of Digital Direct Flat-Panel Mammography Versus an Analog Screen-Film Technique Using a Phantom Model. AJR. 2007;188:399-407.

5. Brazilian Health Ministry. Portaria da Secretaria de Vigilância Sanitária n?453. Diretrizes de Proteção Radiológica em Radiodiagnóstico Médico e Odontológico. Diário Oficial da União, Brasília, 02 de janeiro de 1998.

6. Metz CE. Receiver Operating Characteristic. International Commission on Radiation Units and Measurements. ICRU News. June 7-16, 1997.