Breast Dose Reduction Options During Thoracic CT: Influence of Breast Thickness

Cardiopulmonary Imaging Original Research Revel et al. Breast Dose Reduction in Thoracic CT Cardiopulmonary Imaging Original Research Marie-Pierre Revel 1 Isabelle Fitton 2 Etienne Audureau 3 Joseph Benzakoun 1 Mathieu Lederlin 4 Marie-Laure Chabi 1 Pascal Rousset 5 Revel MP, Fitton I, Audureau E, et al. Keywords: breast dose, CT, radiation dose, radiation savings DOI:10.2214/AJR.14.13255 Received May 21, 2014; accepted after revision July 20, 2014. 1 Department of Radiology, Hotel-Dieu and Cochin Hospital, 27 rue du Fg Saint Jacques, Université Paris Descartes, Sorbonne Paris Cité, 75014 Paris, France. Address correspondence to M. P. Revel (marie-pierre.revel@cch.aphp.fr). 2 Department of Radiology, Pompidou Hospital, Université Paris Descartes, Sorbonne Paris Cité, Paris, France. 3 Department of Biostatistics, Hotel-Dieu Hospital, Université Paris Descartes, Sorbonne Paris Cité, Paris, France. 4 Department of Radiology, CHU Pontchaillou, Université Rennes 1, Rennes, France. 5 Department of Radiology, Hospices Civils de Lyon, Centre Hospitalier Lyon Sud, Université Claude Bernard Lyon 1, Lyon, France. WEB This is a web exclusive article. AJR 2015; 204:W421 W428 0361 803X/15/2044 W421 American Roentgen Ray Society Breast Dose Reduction Options During Thoracic CT: Influence of Breast Thickness OBJECTIVE. Little is known about the effectiveness of dose reduction options according to breast thickness. The purpose of this phantom study was to compare the effects on dose and noise of bismuth shielding versus a low kilovoltage for different breast thicknesses. MATERIALS AND METHODS. CT acquisitions were performed first at 120 kvp (reference acquisition), then at 120 kvp with shielding and at 100 kvp without shielding on a phantom with three different prosthetic breast thicknesses, corresponding to the minimum, median, and maximum values first measured in a sample of 30 female thoracic CT examinations, which were randomly selected. Breast doses were measured with optically stimulated luminescence dosimeters placed on and beneath the prosthetic breast. For noise evaluation, the CT number SDs were measured within six ROIs at increasing depths. RESULTS. Taking into account all breast thicknesses, the average breast dose was reduced by 42.1% with shielding and by 33.0% at 100 kvp (p = 0.009). In-depth noise increased less with shielding (19.0% vs 32.1%, p < 0.0001). For 1-cm breast thickness, the breast dose fell by 46.5% and 29.7% with shielding and 100 kvp, respectively (p = 0.01), and in-depth noise increased by 19.5% and 33.9% (p = 0.01). The corresponding values for 2-cm breast thickness were 38.5% and 30.1%, (p = 0.02) and 16.5% and 33.5% (p = 0.001), whereas those for 4-cm thickness were 40.6% and 40.5% (p = 0.95) and 20.7 and 29.2% (p = 0.02). CONCLUSION. Greater breast dose reduction is achieved by shielding for breast thicknesses less than 4 cm. Regardless of breast thickness, shielding leads to a smaller increase in in-depth noise. T he radiation dose is a concern with thoracic CT because of exposure of the thyroid and breast, two radiosensitive organs. Multiple diagnostic radiographic examinations during childhood and adolescence may increase the risk of breast cancer [1]. Although there are no epidemiologic data for adults, estimates based on studies of atomic bomb survivors suggest that medical imaging may increase the risk of cancer. Smith Bindman et al. [2] reported a mean 13-fold difference between the highest and lowest doses across the 11 most common types of diagnostic CT studies. Taking into account the dose received during coronary CT angiography, they estimated that one in 270 women would develop cancer if CT was performed at age 40 years. Studies of the biologic effects of x-rays offer another approach to estimating this risk. Radiation-induced DNA double-strand breaks have been reported in mammary epi- thelial cells after mammographic screening procedures, especially in high-risk patients. Low and repeated doses lead to greater unrepaired double-strand breaks [3, 4]. Several methods have been proposed to reduce the breast dose, such as bismuth shielding, a globally reduced tube current, and organ-based tube current modulation (OBTCM), a technique developed to lower the CT radiation dose to anterior dose-sensitive organs [5]. OBTCM was found to be the best option in a phantom study of the three approaches [6]. OBTCM can even be used for low-dose chest CT without affecting image quality [7] and producing only a small increase in noise [8]. However, this option is not available on all CT devices. In addition, although OBTCM has been shown to efficiently reduce the breast dose in phantom studies, lateral displacement of the breast in women lying supine for CT examination was not taken into account. A recent report indicated that, in 99% of women lying supine, the angular po- AJR:204, April 2015 W421

Revel et al. sition of breast tissue in the supine position exceeds the limit at which the reduced dose is delivered and partly corresponds to the area where the compensatory increased dose is used [9]. Reducing the tube voltage is another effective option for breast dose reduction [10] independently from breast disposition and is available with all CT devices. The impact of breast size on the effectiveness of the different dose reduction options has not been adequately evaluated. The most important dimension to consider for CT is breast thickness, defined as the distance from the chest wall to the anterior breast tangency line in women lying supine for CT examination. The purpose of this study was to compare the breast dose and image noise on CT acquisitions performed with bismuth shielding versus a low kilovoltage using a phantom with three different prosthetic breast thicknesses. Materials and Methods Phantom and Prosthetic Breasts A tissue-equivalent anthropomorphic torso phantom (IMRT Thorax Phantom Model 002LFC, CIRS) was used for radiation dose and image noise measurement (Fig. 1). The phantom measured 30 cm along the z-axis, 30 cm in the lateral direction, and 20 cm in the anteroposterior direction. It was made of tissue-equivalent epoxy materials and included water-equivalent, lung-equivalent, and bone-equivalent solid rod inserts. To represent patients with different breast sizes, we used prosthetic breasts made of a tissue-equivalent material (wax) placed on the phantom (Fig. 1). The attenuation was close to that of fat, with a mean value of 33 HU ( 39 to 26 HU). Three different prosthetic breast thicknesses were used: 1, 2, and 4 cm, corresponding respectively to the minimum, median, and maximum breast thicknesses measured on 30 thoracic CT examinations (Fig. 2) performed for various clinical indications in women from 17 to 92 years old (mean, 61 years), selected randomly from our PACS. Patient information was anonymized and deidentified before analysis. This retrospective study was approved by the local ethics committee (Comité de Protection des Personnes Ile de France 1), which waived the need for informed consent. CT Acquisitions Reference CT acquisition CT acquisitions were performed on a 64-MDCT scanner (LightSpeed VCT, GE Healthcare) with the following parameters: 0.625-mm collimation; 0.6 second per rotation; pitch, 0.984; 100-mA tube current without dose modulation; 120-kVp tube potential; and reconstruction of 0.625-mm-thick contiguous transverse images using a standard reconstruction algorithm over the full z-extent of the phantom. A scout radiograph was used to assist with the selection of the start and stop locations of the scanning range, which was identical from one CT acquisition to another. Dose-reducing CT acquisitions Two different dose-reducing CT acquisitions were performed: one with tube potential lowered to 100 kvp and no other changes in the CT parameters, and one with the same parameters as the reference CT acquisition using a breast shield made of bismuthimpregnated synthetic rubber on a foam base (AttenuRad CT Breast Shield System, F&L Medical Products) placed above the prosthetic breast. The adult shield (1-mm-thick bismuth, 0.060 mm Pb equivalent, 53 20 cm) was used, allowing the anterior surface of the phantom and the anterior half of its lateral surface to be covered. Each of the three CT acquisitions (reference, 100 kvp, and 120 kvp with shielding) was repeated five times for each of the three prosthetic breast thicknesses (45 acquisitions in total). Radiation Dose Measurement Optically stimulated luminescence dosimeters (nanodot, Landauer) were placed beneath the prosthetic breast and on its anterior surface before each CT acquisition. Measured doses at these two positions (below and above) were averaged to obtain the whole-breast dose. The nanodot dosimeters were read in a microstar reader (Landauer). Each nanodot was used only once, although the system allows full reinitialization. We used screened nanodot dosimeters with ± 5% accuracy and precision in dose measurement. Dose measurements were performed after each of the 45 CT acquisitions. Dose readings were repeated three times per CT acquisition to assess reading reproducibility. Image Noise Measurement Image noise was evaluated by measuring the SD of CT numbers in circular ROIs of 150 mm 2 placed at four different depths in the central part of the phantom (ROIs 1 4), simulating the mediastinum. Noise within the lung of the phantom Fig. 1 Phantom and prosthetic breasts. A, Photograph shows unshielded CT acquisition of phantom with 4-cm-thick prosthetic breasts. B, Photograph shows prosthetic breasts of 4-, 2-, and 1-cm thicknesses made of wax. A B W422 AJR:204, April 2015

Breast Dose Reduction in Thoracic CT TABLE 1: Breast Dose Reduction for All Breast Thicknesses Fig. 2 23-yearold woman with pulmonary embolism suspicion, referred for CT angiography. Measurement of breast thickness on thoracic CT. Breast thickness is defined as distance from chest wall to anterior breast tangency line (double-headed arrow and parallel lines) in supine position during CT examination. When asymmetric, side with largest thickness was taken into account. 120 kvp Without Shielding 100 kvp 120 kvp With Shielding (Reference Acquisition) Below 6.3 ± 0.6 4.2 ± 0.5 32.9 3.7 ± 0.3 41.6 0.004 Above 5.7 ± 1.1 3.8 ± 0.9 33.1 3.3 ± 0.5 42.7 0.035 Average 6.0 ± 0.8 4.0 ± 0.7 33.0 3.5 ± 0.3 42.1 0.009 Note Breast dose reduction was larger with shielding than with 100 kvp option. a 100 kvp vs 120 kvp with shielding. TABLE 2: Breast Dose Reduction According to Each Breast Thickness by Thickness 120 kvp Without Shielding 100 kvp 120 kvp With Shielding (Reference Acquisition) 1 cm Below 6.8 ± 0.3 4.7 ± 0.4 30.3 3.6 ± 0.1 46.9 0.003 Above 6.4 ± 0.4 4.6 ± 0.3 29.1 3.5 ± 0.2 46.2 0.006 Average 6.6 ± 0.3 4.6 ± 0.3 29.7 3.5 ± 0.2 46.5 0.004 2 cm Below 6.5 ± 0.3 4.3 ± 0.3 34.3 3.9 ± 0.2 39.2 0.230 Above 5.5 ± 1.2 4.1 ± 0.1 25.2 3.4 ± 0.4 37.7 0.008 Average 6.0 ± 0.7 4.2 ± 0.2 30.1 3.7 ± 0.2 38.5 0.020 4 cm Below 5.6 ± 0.7 3.7 ± 0.3 34.5 3.5 ± 0.3 37.9 0.485 Above 5.2 ± 1.2 2.8 ± 0.9 46.3 3.0 ± 0.6 43.6 0.809 Average 5.4 ± 0.9 3.3 ± 0.5 40.5 3.2 ± 0.4 40.6 0.953 Note Breast dose reduction was larger with shielding, except for 4-cm breast thickness, for which dose reduction was similar with the two options. The average breast dose with shielding was not significantly influenced by breast thickness (3.5 ± 0.2, 3.7 ± 0.2, and 3.2 ± 0.4, respectively; p = 0.09). a 100 kvp vs 120 kvp with shielding. p a p a was evaluated by placing two additional ROIs laterally, one anteriorly (ROI 6) and another posteriorly (ROI 5) (Fig. 3). Noise was evaluated both overall and by phantom region (superficial noise, ROIs 1 and 2; deep noise, ROIs 3 and 4; and lateral noise, ROIs 5 and 6). Statistical Analysis Results are reported as means (± SD) and as the relative percentage deviation from the reference conditions (120 kvp without shielding). Error bars represent true SD bars. Dose (above, below, and average) and noise (superficial and deep) values from the two dose-reducing CT acquisitions were compared both together and separately for each breast thickness (minimum, median, and maximum) with the two-tailed Student t test for paired data or the Wilcoxon signed rank test, as appropriate. Average breast doses were compared for each breast thickness with the Kruskal-Wallis test. The reproducibility of dose and noise measurements was assessed by computing the intraclass correlation coefficient (ICC) [11]. ICC values of 0.6 0.8 and 0.8 were considered to represent good and excellent reproducibility, respectively. Values for p below 0.05 were considered to denote statistical significance. All statistical analyses were performed using Stata 12.1 software (Stata Corporation). Results Breast Dose Assessment of repeatability For dose reading repeatability (three readings of each nanodot), the ICC was 0.977 (95% CI, 0.961 0.991). For dose measurement repeatability (dose value from each of five similar CT acquisitions) the ICCs were 0.958 (0.913 1000), 0.776 (0.569 0.984), and 0.643 (0.355 0.932) for 1-, 2-, and 4-cm breast thicknesses, respectively. Breast Dose Reduction Considering all the prosthetic breast thicknesses together (Fig. 4), the average breast dose reduction was larger with shielding ( 42.1%) than with a tube potential of 100 kvp ( 33.0%) (p = 0.009). Breast dose reductions at the two positions (below and above the prosthetic breast) were also larger with shielding (Table 1). Taking each breast thickness separately (Fig. 5), the dose reduction was larger with shielding for 1- and 2-cm breast thicknesses (p = 0.004 and p = 0.02), whereas no difference was found between the two protocols for the largest breast thickness (Table 2). The dose measured on the prosthetic breast surface (above) was not significantly different from the dose measured beneath the AJR:204, April 2015 W423

Revel et al. breast, except when shielding was used with the 4-cm-thick prosthetic breast (Table 3). Fig. 3 Image noise measurement on shielded CT acquisition, with 4-cm-thick prosthetic breast phantom. Image noise was evaluated by measuring SD of CT numbers in circular ROIs of 150 mm 2 placed centrally at increasing depths (ROIs 1 4) as well as in mediastinal zone of phantom and laterally in posterior (ROI 5) and anterior (ROI 6) lungs of phantom. This image was acquired at level of rod inserts (unnumbered circles). Note there are no streak artifacts caused by shield. TABLE 3: Comparison of Breast Doses Measured Below and Above Prosthetic Breast by Thickness Acquisition Protocol Below (M ± SD) Above (M ± SD) p 120 kvp without shielding 6.8 ± 0.3 6.4 ± 0.4 0.126 1 cm 100 kvp 4.7 ± 0.4 4.6 ± 0.3 0.12 120 kvp with shielding 3.6 ± 0.1 3.5 ± 0.2 0.17 2 cm 4 cm 120 kvp without shielding 6.5 ± 0.3 5.5 ± 1.2 0.09 100 kvp 4.3 ± 0.3 4.1 ± 0.1 0.49 120 kvp with shielding 3.9 ± 0.2 3.4 ± 0.4 0.116 120 kvp without shielding 5.6 ± 0.7 5.2 ± 1.2 0.336 100 kvp 3.7 ± 0.3 2.8 ± 0.9 0.117 120 kvp with shielding 3.5 ± 0.3 3.0 ± 0.6 0.03 Note Negligible differences were generally found between doses measured below and above the breast, meaning that the superficial dose may be used in vivo as a surrogate for deep dose measurement, except with shielding and the 4-cm breast thickness. In the latter conditions, the superficial breast dose (above) measured close to the shield was slightly lower than that measured under the breast. TABLE 4: Central Noise Increase for All Breast Thicknesses Noise (HU) 120 kvp Without Shielding 100 kvp 120 kvp With Shielding (Reference Acquisition) Superficial 24.4 ± 1.5 31.4 ± 2.4 28.7 31.5 ± 2.3 29.1 0.589 Deep 31.4 ± 1.8 41.5 ± 2.1 32.1 37.4 ± 2.6 19.0 < 0.0001 All 27.9 ± 1.4 36.4 ± 2.1 30.5 34.5 ± 2.2 23.7 < 0.0001 Note There was a smaller increase in central in-depth noise with shielding. a 100 kvp vs 120 kvp with shielding. p a Image Noise The ICC for image noise measurement repeatability was 0.920 (0.885 0.955). Increase in Image Noise Central part of the phantom Image noise was higher with both dose-reducing protocols than with the reference protocol. Noise increased less with shielding than with 100 kvp (23.7% vs 30.5%, p < 0.0001), mainly because of a lesser increase in in-depth noise (19.0% vs 32.1%, p < 0.001). Noise within the two superficial ROIs increased similarly with the two protocols (Fig. 6 and Table 4). Influence of breast thickness on the central noise increase The increase in in-depth noise was larger with the 100 kvp option for all breast thicknesses (Fig. 7), whereas superficial noise increased similarly with the two dose-reduction protocols (Table 5). Increase in Lateral Noise In the phantom lung (ROIs 5 and 6), noise increased similarly with the two protocols (16.3% at 100 kvp and 15.3% with shielding, p = 0.56), although deep noise (ROI 5) increased less with shielding (16.6%, vs 24.9% with 100 kvp, p = 0.03). Artifacts There were no streak artifacts due to the shield on all 15 shielded acquisitions regardless of the breast thickness and the related 1-, 2-, or 4-cm shield-to-chest wall distance. Discussion In this comparative study of two CT breast dose-reduction techniques, shielding was more effective than the use of low kilovoltage when all breast thicknesses were considered together. Shielding was also more effective for the 1- and 2-cm breast thicknesses, whereas no difference was observed for the 4-cm thickness. Although noise increase is less important because it can now be managed by iterative reconstruction techniques, shielding was also associated with a smaller increase in central in-depth noise. The use of a low kilovoltage has been clinically validated as an effective option for global dose reduction during cardiac CT or CT angiography for pulmonary embolism [12 14] because it enables higher arterial enhancement, but its use is largely limited to patients with a body mass index below 25 kg/m 2 [15] or body weight below 75 kg [16]. W424 AJR:204, April 2015

Breast Dose Reduction in Thoracic CT TABLE 5: Central Noise Increase According to Breast Thickness Noise by Thickness 120 kvp Without Shielding 100 kvp 120 kvp With Shielding (Reference Acquisition) (HU) (HU) (HU) Minimum Superficial 23.2 ± 0.9 29.5 ± 1.1 27.2 30.2 ± 0.6 30.2 0.220 Deep 30.3 ± 0.4 40.6 ± 2.7 33.9 36.2 ± 1.7 19.5 0.012 All 26.8 ± 0.4 35.1 ± 1.8 31.0 33.2 ± 0.5 23.9 0.059 Median Superficial 24.1 ± 0.8 30.3 ± 1.2 26.0 30.0 ± 1.6 24.6 0.309 Deep 30.7 ± 1.4 40.9 ± 1.8 33.5 35.7 ± 1.1 16.5 0.001 All 27.4 ± 0.6 35.6 ± 1.4 29.9 32.8 ± 1.2 19.7 0.001 Maximum Superficial 25.8 ± 1.6 34.2 ± 1.0 32.7 34.3 ± 0.4 33.1 0.880 Deep 33.4 ± 1.3 43.1 ± 0.9 29.2 40.3 ± 1.9 20.7 0.020 All 29.6 ± 1.1 38.7 ± 0.9 30.7 37.3 ± 0.8 26.0 0.016 Note Breast thickness may influence the increase in noise because shield is placed on prosthetic breast and distance from chest wall to shield depends on prosthetic breast thickness (1, 2, or 4 cm). There were no significant differences between the two options (shielding vs 100 kvp) regarding the increase in superficial noise, whereas deep noise was increased less by shielding regardless of breast thickness. a 100 kvp vs 120 kvp with shielding. 8 7 6 5 4 3 2 1 0 Under Above Average Position Fig. 4 Breast dose considering all breast thicknesses together. Graphs of reference protocol of 120 kvp without shielding (dark blue) and two dose-reducing options of 100 kvp (medium blue) and 120 kvp with shielding (light blue) show data from Table 1. Average breast dose was not significantly different from superficial (above) and deep (under) doses and was lower with shielding. Error bars represent SD of measurements. p a These latter studies did not specifically address breast dose reduction but considered the entire radiation dose by evaluating the dose-length product or the effective radiation dose, whereas our study protocol consisted of direct breast dose measurements. Bismuth shielding has been shown in several studies to reduce the breast dose in vivo without impairing image quality. Hurwitz et al. [17] reported a 55% dose reduction on CT angiography, Yilmaz et al. [18] reported 40% dose savings during routine chest CT in adults, and Fricke et al. [19] reported a 29% breast dose reduction in a pediatric population. Shielding combined with ECG-triggered tube current modulation yielded breast dose reductions of 53 63% during cardiac CT [20]. In contrast, phantom studies have not supported the use of breast shielding because of an increase in image noise and artifacts together with changes in image attenuation characteristics [21]. Likewise, Vollmar et al. [22] reported that noise increased by up to 40%, and Wang et al. [6] observed beam and streak artifacts with shielding as well as more noise, problems that were not observed with OBTCM. These discrepancies are largely explained by differences in the definition of image quality (qualitative or quantitative) and by how the shield is placed with respect to a critical point namely the distance between the shield and the anterior aspect of the phantom. Our results are in line with those of a study by Kalra et al. [23] evaluating the effects of the shield-to-surface distance. We observed no streak artifacts because the bismuth shield overlaid the prosthetic breast and was thus at least 1-cm distant corresponding to the smallest prosthetic breast thickness used in our study from the phantom s anterior surface. Streak artifacts have not been reported for shield-to-surface distances more than 1 cm. We found that shielding induced a smaller increase in central in-depth noise for all three simulated breast thicknesses, which was in keeping with the report by Kalra et al. [23] of a smaller increase in noise in the posterior regions of the phantoms, whatever the shield-to-surface distance. Using Monte Carlo simulation, Geleijns et al. [24] estimated that lowering the tube current to obtain similar image noise would reduce the breast dose more efficiently than the use of breast shields, whereas Foley et al. [25] reached the opposite conclusion in another recent phantom study. AJR:204, April 2015 W425

Revel et al. 8 7 6 5 4 3 2 1 0 1 cm 2 cm 4 cm Thickness Fig. 5 Breast dose for each breast thickness. Graphs of reference protocol of 120 kvp without shielding (dark blue) and two dose-reducing options of 100 kvp (medium blue) and 120 kvp with shielding (light blue) show data from Table 2. Breast dose reduction was larger with shielding for minimum and median thicknesses (1 and 2 cm), whereas it was not significantly different between two options for largest breast thickness. Image Noise (HU) 50 45 40 35 30 25 20 15 10 5 0 The observed effects of dose-reduction options are highly dependent on how the breast dose is calculated. In most studies, Monte Carlo based computer programs for dose estimates have been used, which is not equivalent to obtaining true breast dose measurements. We made direct dose measurements, both superficially and beneath the prosthetic breasts, to better evaluate the whole-breast dose. This enabled us to measure beam attenuation inside the breast and to verify whether the surface dose is an acceptable surrogate for CT breast dose measurement in vivo, although this was not the main objective of our study. We found that the superficial dose value was not significantly different from the deep dose during the reference and 100-kVp protocols, whereas this was not the case with shielding, at least for the 4-cm-thick prosthetic breast. Kim et al. [7] also made direct breast dose measurements and found that with OBTCM and shielding, respectively, the surface breast dose was reduced by 20% and 16% and the deep breast dose by 18% and 28%. Differences in the CT equipment used in all these studies may partly explain the controversy surrounding shielding [26]. Depending on the CT equipment, shielding may interact with the automatic exposure control technique. With automated ROI 1 ROI 2 ROI 3 ROI 4 ROI 5 ROI 6 Superficial Deep Fig. 6 Image noise with reference protocol of 120 kvp without shielding (dark blue) and two dose-reducing options of 100 kvp (medium blue) and 120 kvp with shielding (light blue). Graphs show ROIs 1 6 and superficial and deep noise. With shielding, noise increase was smaller in posterior ROIs (deep noise). ROIs 1 and 6 were placed on anterior regions. W426 AJR:204, April 2015

Breast Dose Reduction in Thoracic CT technologies that modulate the tube current in the z-axis on the basis of the topogram, it is important to place the shield only after obtaining the scout image; otherwise the automated exposure control may compensate for the presence of the shield by increasing the tube current [27]. In a phantom study using a 16-MDCT scanner (Light- Speed, GE Healthcare), the effective dose was reduced by 35% when the shield was placed after the scout view acquisition and by 20% when it was placed before the scout acquisition [28]. Other technologies adjust the tube current on the basis of direct measurement of the radiation intensity during acquisition. Thus, there is a risk that the dose will be increased due to the high attenuation of the shield. It is generally agreed that the results may be unpredictable if real-time x-y dose modulation is used to adjust exposure during acquisition [27]. We used a constant tube current and did not evaluate the interaction of the shield with the automatic exposure control of our equipment to avoid potential confounding factors. To our knowledge the influence of breast size on the breast dose with and without Image Noise (HU) 50 45 40 35 30 25 20 15 10 5 0 shielding has previously been investigated in only one series, using cardiac CT [29]. The mean breast dose received by the large breast phantom was smaller than that of the medium-sized breast phantom, and the authors attributed this to the fact that, with the large-breast phantom, a larger proportion of the breast was outside the field of the cardiac CT examination, which is not the case during standard thoracic CT. In our study, we explain the larger dose for thinner prosthetic breast by the fact x-rays are less attenuated. There are some limitations in our study. We only evaluated three breast thicknesses and did not assess the effects of shielding for breast thicknesses more than 4 cm because this was the largest distance between the chest wall and the anterior breast tangency line measured in vivo on 30 thoracic CT examinations. Although higher values may be observed in larger study populations, this would not affect our main finding that is, the higher efficacy of shielding less than 4-cm breast thickness. We only used one CT device, but similar results are likely to be obtained with other devices because we used a constant tube current rather than automated modulation, the modalities of which differ from one manufacturer to another. The final limitation is that we did not evaluate the impact of breast density because all prosthetic breasts used in our study had the same attenuation value, close to that of fat. However, this does not change our main result, which is that for a given breast density, shielding more effectively reduces breast dose than kilovoltage reduction for small breast thicknesses. The relationship between breast thickness and density should be evaluated on a dedicated study. Even though we might suspect larger breasts to be fatty, patient age must be considered and the former assumption may not be true in younger patients. In conclusion, breast thickness influences the effectiveness of breast dose reduction options. Shielding provides a larger reduction than a tube voltage of 100 kvp for breast thicknesses less than 4 cm. This suggests shielding is an important option, especially for unenhanced CT examinations in which the higher contrast enhancement allowed by low kilovoltage is not a requirement. Superficial Deep Superficial Deep Superficial Deep 1 cm 2 cm 4 cm Thickness Fig. 7 Increase in central noise according to breast thickness and related shield-to-surface distance with reference protocol of 120 kvp without shielding (dark blue) and two dose-reducing options of 100 kvp (medium blue) and 120 kvp with shielding (light blue). Graphs show image noise compared with 1-, 2-, and 4-cm-thick phantoms. AJR:204, April 2015 W427

Revel et al. References 1. Doody MM, Lonstein JE, Stovall M, Hacker DG, the female breast as a result of spectral optimisation for high-contrast thoracic CT imaging: a mental and clinical data. AJR 2003; 180:407 411 20. Midgley SM, Einsiedel PF, Langenberg F, Lui Luckyanov N, Land CE. Breast cancer mortality phantom study. Br J Radiol 2009; 82:920 929 EH, Heinze SB. Assessment of patient dose and after diagnostic radiography: findings from the 11. Shrout PE, Fleiss JL. Intraclass correlations: uses image quality for cardiac CT with breast shields. U.S. Scoliosis Cohort Study. Spine 2000; 25:2052 in assessing rater reliability. Psychol Bull 1979; Radiat Prot Dosimetry 2012; 151:463 468 2063 86:420 428 21. Goldin D, Emerson S, Farah MC, et al. Shielding 2. Smith-Bindman R, Lipson J, Marcus R, et al. Radiation dose associated with common computed tomography examinations and the associated lifetime attributable risk of cancer. Arch Intern Med 2009; 169:2078 2086 3. Colin C, Devic C, Noel A, et al. DNA doublestrand breaks induced by mammographic screening procedures in human mammary epithelial cells. Int J Radiat Oncol Biol 2011; 87:1103 1112 4. Colin C, Foray N. DNA damage induced by mammography in high family risk patients: only one single view in screening. Breast 2012; 21:409 410 5. Duan X, Wang J, Christner JA, Leng S, Grant KL, McCollough CH. Dose reduction to anterior surfaces with organ-based tube-current modulation: evaluation of performance in a phantom study. AJR 2011; 197:689 695 6. Wang J, Duan X, Christner JA, Leng S, Yu L, Mc- Collough CH. Radiation dose reduction to the breast in thoracic CT: comparison of bismuth shielding, organ-based tube current modulation, and use of a globally decreased tube current. Med Phys 2011; 38:6084 6092 7. Kim YK, Sung YM, Choi JH, Kim EY, Kim HS. Reduced radiation exposure of the female breast during low-dose chest CT using organ-based tube current modulation and a bismuth shield: comparison of image quality and radiation dose. AJR 2013; 200:537 544 8. Matsubara K, Sugai M, Toyoda A, et al. Assessment of an organ-based tube current modulation in thoracic computed tomography. J Appl Clin Med Phys 2012; 13:3731 9. Taylor S, Litmanovich DE, Shahrzad M, Bankier AA, Gevenois PA, Tack D. Organ-based tube current modulation: are women s breasts positioned in the reduced-dose zone? Radiology 2015; 274:260 266 10. Vollmar SV, Kalender WA. Reduction of dose to 12. Schueller-Weidekamm C, Schaefer-Prokop CM, Weber M, Herold CJ, Prokop M. CT angiography of pulmonary arteries to detect pulmonary embolism: improvement of vascular enhancement with low kilovoltage settings. Radiology 2006; 241:899 907 13. Heyer CM, Mohr PS, Lemburg SP, Peters SA, Nicolas V. Image quality and radiation exposure at pulmonary CT angiography with 100- or 120- kvp protocol: prospective randomized study. Radiology 2007; 245:577 583 14. Halliburton SS, Abbara S, Chen MY, et al. SCCT guidelines on radiation dose and dose-optimization strategies in cardiovascular CT. J Cardiovasc Comput Tomogr 2011; 5:198 224 15. Feuchtner GM, Jodocy D, Klauser A, et al. Radiation dose reduction by using 100-kV tube voltage in cardiac 64-slice computed tomography: a comparative study. Eur J Radiol 2010; 75:e51 e56 16. Sigal-Cinqualbre AB, Hennequin R, Abada HT, Chen X, Paul JF. Low-kilovoltage multidetector row chest CT in adults: feasibility and effect on image quality and iodine dose. Radiology 2004; 231:169 174 17. Hurwitz LM, Yoshizumi TT, Goodman PC, et al. Radiation dose savings for adult pulmonary embolus 64-MDCT using bismuth breast shields, lower peak kilovoltage, and automatic tube current modulation. AJR 2009; 192:244 253 18. Yilmaz MH, Albayram S, Yasar D. al. Female breast radiation exposure during thorax multidetector computed tomography and the effectiveness of bismuth breast shield to reduce breast radiation dose. J Comput Assist Tomogr 2007; 31:138 142 19. Fricke BL, Donnelly LF, Frush DP, et al. In-plane bismuth breast shields for pediatric CT: effects on radiation dose and image quality using experi- artificially increases the attenuation of water: study of CT gradient attenuation induced by shielding (CT GAINS). Acad Radiol 2013; 20:231 237 22. Vollmar SV, Kalender WA. Reduction of dose to the female breast in thoracic CT: a comparison of standard-protocol, bismuth-shielded, partial and tube-current-modulated CT examinations. Eur Radiol 2008; 18:1674 1682 23. Kalra MK, Dang P, Singh S, Saini S, Shepard JA. In-plane shielding for CT: effect of off-centering, automatic exposure control and shield-to-surface distance. Korean J Radiol 2009; 10:156 163 24. Geleijns J, Salvado Artells M, Veldkamp WJ, Lopez Tortosa M, Calzado Cantera A. Quantitative assessment of selective in-plane shielding of tissues in computed tomography through evaluation of absorbed dose and image quality. Eur Radiol 2006; 16:2334 2340 25. Foley SJ, McEntee MF, Rainford LA. An evaluation of in-plane shields during thoracic CT. Radiat Prot Dosimetry 2013; 155:439 450 26. McCollough CH, Wang J, Gould RG, Orton CG. Point/counterpoint: the use of bismuth breast shields for CT should be discouraged. Med Phys 2012; 39:2321 2324 27. Colletti PM, Micheli OA, Lee KH. To shield or not to shield: application of bismuth breast shields. AJR 2013; 200:503 507 28. Coursey C, Frush DP, Yoshizumi T, Toncheva G, Nguyen G, Greenberg SB. Pediatric chest MDCT using tube current modulation: effect on radiation dose with breast shielding. AJR 2008; 190:[web] W54 W61 29. Einstein AJ, Elliston CD, Groves DW, et al. Effect of bismuth breast shielding on radiation dose and image quality in coronary CT angiography. J Nucl Cardiol 2012; 19:100 108 W428 AJR:204, April 2015