Mammograms Obtained with Rhodium vs Molybdenum Anodes: Contrast and Dose Differences

1313 C. Kimme-Smith1 J. Wang2 N. DeBruhl1 M. Basic2 L. W. Bassett1 Received October 27, 1 993; accepted after revision January 25, 1994. Presented at the annual meeting of the American Roentgen Ray Society, San Francisco, April 1993. 1The Iris Cantor Center for Breast Imaging, University of California, Los Angeles, School of Medicine, 10833 Le Conte Ave., Los Angeles, CA 90024. Address correspondence to C. Kimme- Smith. 2Biomedical Physics Graduate Program, University of Califomia, Los Angeles, School of Medicine, Los Angeles, CA 90024. 0361-803X/94/1626-1 313 American Roentgen Ray Society Mammograms Obtained with Rhodium vs Molybdenum Anodes: Contrast and Dose Differences OBJECTIVE. A mammography unit with both a molybdenum anode and a rhodium anode, filtered with molybdenum and rhodium, respectively, was evaluated to determine which types of women would benefit from the dose savings of the rhodium combination despite some loss of contrast. SUBJECTS AND MATERIALS. In 100 women, the molybdenum anode and molybdenum filtration (Mo/Mo) were used to obtain mammograms of the right breast, and the rhodium anode and rhodium filtration (Rh/Rh) were used for mammograms of the left breast. All mammograms were obtained at 26 kvp. All milliampere-second values used to radiograph the breasts of these women were recorded. Mammograms of 54 women (30 with previous mammograms available), representing the four types of breasts as defined by the American College of Radiology, were interpreted by three radiologists. Each mammogram was assigned a grade for breast type, preference (Rh/Rh, Mo/Mo, or previous mammograms), contrast, and sharpness. RESULTS. Overall, mammograms obtained by using the Mo/Mo combination were preferred. However, for images of types 3 and 4 breasts, Rh/Rh was preferred twice as often as it had been for mammograms of types 1 and 2 breasts. The mean glandular dose for all breast types when the Rh/Rh combination was used was 42% of the dose used for the Mo/Mo combination. For a 6-cm-thick dense breast, the Rh/Rh combination required 40% of the dose required for the Mo/Mo combination. CONCLUSION. Mammograms obtained with the Rh/Rh combination carried an overall decrease in contrast and mean glandular dose. However, for young women and some women with large dense breasts, the Rh/Rh mammograms were equivalent to or better than the mammograms obtained with the Mo/Mo combination. Effective use of Rh/Rh units requires careful selection of women based on age or the amount of glandular tissue seen on previous mammograms. AJR 1994;162:1313-1317 For at least 20 years, physicists have been modeling and predicting the optimal energy spectrum for mammography. The criteria used have varied from signal-tonoise ratio vs dose, to contrast, to quality factors that are combinations of more sophisticated measurements of noise, dose, resolution, and contrast [1-4]. All have agreed that for breasts compressed to 2 cm, a molybdenum anode with molybdenum filtration (Mo/Mo) is preferable, and for compressed breasts 6 cm on thicker, combinations of anodes and filters that include rhodium will require a smaller dose of radiation and can give comparable image quality. Some of these tests have been simulations based on spectral analysis [1, 3] or experimental exposures of phantoms and wedges [4]. Others have used mastectomy samples to test for contrast [2]. The earliest experiments included kilovoltages and tube filtration that are no longer relevant now that screen-film mammography is used much more often than xeronadiography. The range of kilovoltages needed for optimal contrast can be predicted from measurements of the attenuation coefficient of breast tissue [5]. Measurements of mean glandular dose have been refined for mammograms obtained with Mo/Mo [6] and recently for those obtained with

1314 KIMME-SMITH ET AL. AJR:162, June 1994 molybdenum anodes and rhodium filtration and rhodium anodes and rhodium filtration (Rh/Rh) [7]. Much of this early work was not tested clinically because mammographic equipment filtered with nonmolybdenum elements was not available or was not clinically useful. Recently, General Electric (Waukesha, WI) introduced a new unit, the DMR, that has both molybdenum and rhodium tracks on the anode and separate cathodes for each anode track. For screen-film energies, either molybdenum on rhodium filters can be selected (the rhodium anode cannot be filtered with molybdenum). Because phantoms and simulations were used in previous experiments on exotic combinations of anodes and filters, we designed a clinical trial to investigate the dose of radiation received by the patient and the contrast and sharpness of the images when the DMR unit was used to obtain mammograms in a varied group of women. Although it was clean that many of the mammograms obtained with the rhodium anode were inferior in contrast to those obtained with the molybdenum anode, some breasts seemed to be imaged equally well with both systems. In order to take advantage of the dose savings obtained with the Rh/Rh system, this study was designed to quantify the type of breasts that could be imaged successfully with the Rh/Rh system. Subjects and Methods Tests verified that for the Mo/Mo combination, focal spot, kilovoltage, milliroentgens/miiliampere-seconds, half-value layer, compression, imaging characteristics, and dose of radiation with the DMR unit were comparable to those with the GE 600T unit that the DMR unit replaced. For the Rh/Rh combination, invasive measurements of kilovoltage verified that the same circuitry supplied voltage to the two anode tracks equally. Noninvasive measurements were not comparable because noninvasive meters are not available for the Rh/Rh unit. The half-value layer at 26 kvp for the Rh/Rh combination was 0.38 mm of aluminum, compared with 0.33 mm of aluminum for the Mo/Mo combination. Focal spot measurements with a 0.01-mm slit showed that the two cathodes were comparable. The large focal spots were 0.45 x 0.51 mm for the molybdenum anode and 0.46 x 0.62 mm for the rhodium anode at the chest wail. Use of the American College of Radiology (ACR) mammography accreditation phantom necessitated using 26 kvp for clinical testing, because when the Rh/Rh combination was used, images of the phantom obtained at higher kilovoltages were not acceptable. At 26 kvp, the dose of radiation to the ACR phantom was 94.4 mrad (0.944 mgy) when the Mo/Mo combination was used and 45.8 mrad (0.458 mgy) when the Rh/Rh combination was used. From August to October 1992, after acceptance testing, the DMR unit was used to obtain images in 1 00 of 350 women who had screening mammography at our center. Kodak Mm A medium screens and Kodak Mm R E film developed for 44 sec (extended cycle) were used for all mammograms. No selection criteria were applied during the 3 weeks of data collection. For each woman, the Mo/Mo combination at 26 kvp was used for mammograms of two views of the right breast and the Rh/Rh combination at 26 kvp was used for mammograms of two views of the left breast. This sequence was not randomized because an eanlier, randomized trial had numerous technologist errors so that the Rh/Rh images were not always correctly identified. Milliampere-seconds were recorded for both breasts and both views so that mean glandular dose of radiation could be calculated. The mammograms were interpreted as usual by one of five board-certified radiologists. Two months later, the first sequential 34 images were assembled into a test set, and each set of films was evaluated independently by three of the five radiologists. Four months later, 20 additional cases were selected from the remaining 66 cases, on the basis of the amount of radiation received. This selection was made because more than half (65%) of the first 35 patients were graded as having almost entirely fatty breast tissue (ACR type 1) or fatty tissue with scattered fibrogiandular tissue (ACR type 2). The selection of 20 additional cases increased the number of type 3 breasts (heterogeneously dense) in our test set. Nineteen of the examinations revealed caicifications, but no biopsies were done because the calcifications were scattered or appeared benign. Only one woman with implants was included in the study, and no patients were included who had had surgery that would affect the symmetry of the comparison of the right vs the left breast. Of the 54 women included in this study, previous mammograms that had been obtained with an Mo/Mo system operating at 26 kvp were available for 30. These previous mammograms were included in this study to act as a control on the GE DMA equipment. They represented mammograms obtained with conventional equipment. The most recent of these mammograms and the mammograms obtained in our study were mounted on a high-intensity viewer so that mammograms of only one patient were seen at a time and images of the night and the left breasts could be compared. The radiologists who participated in our study knew that the Mo/ Mo combination was used for images of the right breast and the Rh/ Rh combination for images of the left breast. Each radiologist graded the contrast and sharpness of each mammogram on a scale of 1 to 3, with 1 = unacceptable and 3 = excellent. The radiologists also classified the breast tissue according to ACR criteria and selected which method gave preferred results. They also were allowed to have no preference, in which case they specified that images of both right and left breasts were satisfactory (Figs. 1-3). Doses of radiation were calculated from measurements of miiliampore-seconds and from miiliroentgens per milliampere-seconds (milliroentgens/miliiampere-seconds linearity was within 2%). Recently Fig. 1.-A and B, Mammograms of right (A) and left (B) breasts obtamed compressed to 6 cm. Breasts were classified as type 3 according to American College of Radiology criteria. Two radiologists preferred Rh/Rh (B) over Mo/Mo (A) image. A third radiologist found both types of images equally satisfactory. Dose of radiation for Rh/Rh image was 40% of that for Mo/Mo Image.

AJR:162, June 1994 RHODIUM VS MOLYBDENUM ANODES FOR MAMMOGRAPHY 1315 Results Fig. 2.-A and B, Mammograms of right (A) and left (B) breasts obtamed classified as type 2 according to American College of Radiology criteria. Two radiologists preferred Mo/Mo Image, whereas a third radiologist found both types of images equally satisfactory. Fig. 3.-A and B, Mammograms of right (A) and left (B) breasts obtamed classified as type 3 according to American College of Radiology criteria. Breasts were compressed to 4 cm. Unlike mammograms In Fig. 1, all three radiologists preferred Mo/Mo system (A) over Rh/Rh system (B). Because of differences In breast size between breasts In Figs. 1 and 3, dose savings were less for these breasts. Rh/Rh dose of radiation was 45% of Mo/Mo dose. developed tables were used to convert skin exposure to mean gianduian dose [7]. Regression analysis was performed for craniocaudal and mediolateral oblique views. Optical density was measured in fatty tissue near the chest wail for both views. Mammograms obtained with the Mo/Mo combination had an average optical density of 1.58 for either view. For mammognams obtained with the Rh/Rh combination, the average optical density was 1.47 for craniocaudal views and 1.45 for mediolateral oblique views. Table 1 lists the average scones assigned to the films for contrast and sharpness. The most experienced radiologist graded the Rh/Rh images unacceptable with respect to contrast for 13 (24%) of 54 cases. Note that breast types were determined by each radiologist, so some of the breasts classified as type 2 by one radiologist were classified as type 3 by another radiologist. Disparities in breast type occurred in four of the 108 classifications made by the least experienced of the three radiologists who examined the images. This variation is remarkably small, because if the classifications made by the most experienced radiologist were used for all the radiologists instead of each radiologist s classifications being used, changes from the scones in Table 1 would vary by less than 0.1 5. This variation is less than the between-interpreter standard deviation of 0.25 for grading contrast. Analysis of variance and modified t-statistics with Bonfemroni correction for multiple comparisons [8] of the results in Table 1 showed that differences between mammograms obtained with the Mo/Mo combination, mammograms obtained with the Rh/ Rh combination, and previous mammograms were significant (p <.001 ) for contrast ton all breast types (Table 1). Differences between Rh/Rh mammograms and previous mammograms TABLE 1 : Scores of Three Radiologists Evaluating Mo/Mo, Rh/Rh, and Previous Mammograms for Contrast and Sharpness Characteristics/Group of Mammograms Mean Score (±SD ) by Breast Type Types 1 and 2a Types 3 and 4b Contrast Mo/Mo 2.7 ± 0.4 2.6 ± 0.5 Rh/Rh 2.0 ± 0.3 2.0 ± 0.4 Previous 2.2 ± 0.5 2.1 ± 0.4 Sharpness Mo/Mo 2.4 ± 0.5 2.3 ± 0.5 Rh/Rh 2.0 ± 0.4 2.1 ± 0.5 Previous 2.1 ± 0.3 2.2 ± 0.5 Note-The type of breast tissue was determined on the basis of criteria of the American College of Radiology (ACR). On average, 25 breasts were graded ACR breast type 1 or 2, and 21 breasts were graded ACR type 3 or 4. Contrast and sharpness were graded on a scale of 1-3 (1 = unacceptable, 3 = outstanding). Mo/Mo = mammograms obtained with a molybdenum anode and molybdenum filtration, Rh/Rh = mammograms obtained with a rhodium anode and rhodium filtration. Previous mammograms were obtained (in 30 women) with a Mo/Mo system different from the GE DMA unit used for the current mammograms. All systems operated at 26 kvp. afor contrast, p <.001 for Mo/Mo vs Rh/Rh and for Mo/Mo vs previous mammograms; p <.01 for Rh/Rh vs previous mammograms. For sharpness, p <.001 for Mo/Mo vs Rh/Rh and for Mo/Mo vs previous mammograms; Rh/ Rh vs previous mammograms = NS. bfor contrast, p <.001 for Mo/Mo vs Rh/Rh and for Mo/Mo vs previous mammograms; Rh/Rh vs previous mammograms = NS. For sharpness, Mo/ Mo vs Rh/Rh, Mo/Mo vs previous mammograms, and Rh/Rh vs previous mammograms = NS.

1316 KIMME-SMITH ET AL. AJR:162, June 1994 were significant only with respect to contrast for breasts of types 1 and 2 (p <.01 ). This implies that the Rh/Rh mammograms were equivalent in contrast to previous Mo/Mo mammograms for breasts of types 3 and 4. For sharpness, differences were significant only for breasts of types 1 and 2 for Mo/Mo images vs Rh/Rh images and Mo/Mo mammograms vs previous mammograms (p <.001 ). Thus, mammograms obtained by using the Mo/Mo combination had better contrast than both Rh/Rh mammograms and previous mammograms for all breast types. This significance also holds when the results for the least experienced radiologist are deleted. When this radiologist s results are excluded, the diffenence in sharpness for Mo/Mo mammograms vs Rh/Rh mammograms is also significant (p < 01 ) for breasts of types 3 and 4. Figure 4 summarizes the preferences of the three nadiologists who interpreted the mammograms. Preferences were considered equal if the Mo/Mo and Rh/Rh mammograms were judged to be equivalent. In two cases, one radiologist preferred the previous mammograms, and these results were excluded from Figure 4. The few type 1 and type 4 breasts were grouped with type 2 and type 3 breasts, respectively. This figure shows that, although mammograms obtained by using the Mo/Mo combination were preferred for more cases than images obtained with the Rh/Rh systern, this preference declined to 60% for mammograms of dense breasts (types 3 and 4). This is important when mean glandular doses associated with the Rh/Rh combination are compared with those associated with the Mo/Mo combination. Figure 5 shows the overall regression analysis of the mean glandular dose of radiation for all the craniocaudal and mediolatenal oblique views. For craniocaudal views, the average dose (DA) for Rh/Rh mammograms is given by DR = 6.6 + 0.43 DM (r2 =.79); for mediolateral oblique views, DR 12.0 + 0.41 DM (r2 =.86). DM is the average dose associated with Mo/Mo mammograms and r is the coefficient of correlation. When the Rh/Rh combination is used, the total dose savings for all types of breasts is 58% of the typc3. Equal #{149}Rh 0 Mo Fig. 4.-Graph shows percent preferences of radiologists for mammograms of types 2 and 3 breasts obtained by using a rhodium anode and rhodium filtration (Rh) or a molybdenum anode and molybdenum flitration (Mo). Equal represents those instances when the two types of images were judged to be equivalent. Mammograms of two breasts graded as type 1 were included with type 2, whereas mammograms of three breasts graded type 4 were Included with type 3. I 1:: TT: 8O, :: a -r 40 / 1 3O} I :: mean glandular dose for Mo/Mo mammograms. When breasts of types 1 and 2 are compared with breasts of types 3 and 4 (using the most experienced radiologist s grades to avoid discrepancies in grading among interpreters), the dose is reduced to 42% for breasts with less fibroglandulan tissue and 43% for dense bneasts. Discussion Our results verify that a loss in image contrast occurs and the dose of radiation is lower when the Rh/Rh combination instead of the Mo/Mo combination is used for mammography [2]. They also show that in some women who have breasts with more fibnoglandulan tissue, mammograms obtained with the Rh/Rh combination are equal to mammograms obtained with the Mo/Mo combination and that the Rh/Rh combination I 00 200 300 400 A Mo (mrad) Mo (mrad) B Fig. 5.-A and B, Doses of radiation plotted for cranlocaudal (A) and medlolateral oblique (B) mammograms obtained by using a rhodium anode and rhodium filtration (Rh) or a molybdenum anode and molybdenum filtration (Mo).

AJR:162, June 1994 RHODIUM VS MOLYBDENUM ANODES FOR MAMMOGRAPHY 1317 can even be preferred by experienced radiologists so long as 26 kvp is used (Fig. 4). The problem of selecting the most appropriate combination of target and filter for each woman has been addressed by the manufacturer of the unit tested. On this unit, an automatic mode presamples the breast with a short burst of radiation and then the kilovoltage, anode, and filtration are selected. We did not use this automatic selection because our interest was in the larger question of evaluating contrast, sharpness, and dose for Mo/Mo vs Rh/Rh when operating at the same kilovoltage. Table 1 shows an overall improvement in both contrast and sharpness between mammograms obtained with the Mo/Mo combination in this study and previous mammograms obtained with a similar combination. This cannot be attributed entirely to the new equipment. It is more likely a result of improved image-receptor contrast. The film used has improved in contrast and Dmax value in the preceding 2 years. This may increase acceptance of using the Rh/Rh combination to obtain images. Several manufacturers have developed films that provide greaten contrast than Kodak Mm R films do. The use of these films with an Rh/Rh system could restore subject contrast lost with this system. In addition to reducing the dose of radiation, use of the Rh/Rh combination can also reduce the image noise associated with very low dose techniques. Currently, dose is reduced by increasing the light photons emitted by those X- nay photons that exit the breast on by increasing the speed of the film. With the Rh/Rh combination, the number of photons absorbed by breast tissue is decreased. By maintaining the same number of exit photons, the same quantum mottle will occur for an Rh/Rh mammogram as for the Mo/Mo mammogram, but the dose of radiation will be reduced. We originally included noise evaluation in our study, but results for the first 20 cases were so uniform that noise evaluation was eliminated. However, if the number of exit photons is increased, and a slower image receptor is used, quantum mottle can be reduced for Rh/Rh images even while dose is slightly reduced compared with the Mo/Mo combination. The number of women who would benefit from Rh/Rh imaging is limited. In our study, Rh/Rh mammograms were clearly better than Mo/Mo mammograms in only eight of 54 women. In 14 other women, use of the Rh/Rh combination provided no loss of image quality and the dose of radiation was reduced. It would seem that a technique that is suitable for only 41 % (22/54) of patients who have mammography and of positive benefit to only 1 5% (8/54) has limited application. However, the women for whom this technique is most suitable are the ones least served by conventional Mo/Mo mammography. A woman with lange or dense breasts will receive 160-400% of the dose of radiation received by the ACR mammography breast phantom when an Mo/Mo system is used to obtain mammograms. Women with dense breasts tend to be premenopausal or less than 60 years old, so radiation to the breast generally is of more concern. Therefore, it is reasonable to protect these women by reducing the dose of radiation they receive during mammography and by using a technique in which X-rays penetrate glandu- Ian tissue more efficiently, creating a diagnostic mammogram, not one limited by dense fibroglandulan tissue. Our results can be used to select these patients. In this study, we found that in 40% of the patients with type 3 or 4 breasts, Rh/Rh mammograms were preferred or considered to be equivalent to Mo/Mo mammograms. More than half of these patients were less than 55 years old. A strategy for selecting women who might benefit from the Rh/Rh combination would be to select all women whose previous mammograms show their breasts are type 3 on 4 and who also have breasts that cannot be compressed to less than a 6-cm thickness. In addition, women less than 50 years old might benefit from the lower dose of radiation associated with the Rh/Rh combination. The similar results for patients with breasts of types 3 and 4 with respect to film sharpness (Table 1) allow some latitude when selecting women for an Rh/Rh study. These results should reassure radiologists that even if contrast is less on Rh/Rh mammograms, the ability to detect calcifications will not be affected. As more women return for mammography, so that breast type is known, the intelligent use of Rh/Rh equipment will be facilitated. Furthermore, if the mean glandular dose of radiation to the breast in young women can be reduced 50%, the cost-benefit ratio might allow us to screen those younger women who are at higher risk for breast cancer. REFERENCES 1. Johnson GA, O Foghludha F. An experimental trans-molybdenum tube for mammography. Radiology 1976:127:511-516 2. Jennings RJ, Eastgate RJ, Seedband MP, Ergun DL. Optimal x-ray spectra for screen-film mammography. Med Phys 19818:629-639 3. Beaman SA, Lillicrap SC. Optimal x-ray spectra for mammography. Phys Med Biol1982;10:1209-1220 4. Desponds L, Depeursinge C, Grecescu M, Hessler C, Samiri A, Valley JF. Influence of anode and filter material on image quality and glandular dose for screen-film mammography. Phys Med Biol 1 991 36:116-1182 5. Johns PC, Yaffe MJ. X-ray characterization of normal and neoplastic breast tissue. Phys Med Biol 1987;32:675-695 6. wu X, Bames GT, Tucker DM. Spectral dependence of glandular tissue dose in screen-film mammography. Radiology 1 991 ; 179:143-148 7. Wu X, Gingold EL, Bames GT, Tucker DM. Glandular tissue dose in rhodium target/rhodium filter mammography. Med Phys 1993;20:880 8. Wallenstein 5, zucker CL, Fleiss JL. Some statistical methods useful in circulation research. Circ Res 1980;47:1-9