Assessment of scatter radiation dose and absorbed doses in eye lens and thyroid gland during digital breast tomosynthesis

Abstract Digital breast tomosynthesis (DBT) is an alternative tool for breast cancer screening; however, the magnitude of peripheral organs dose is not well known. This study aimed to measure scattered dose and estimate organ dose during mammography under conventional (CM) and Tomo (TM) modes in a specific DBT system. Optically stimulated luminescence dosimeters (OSLDs), whose responses were corrected using a parallel‐plate ionization chamber, were pasted on the surface of custom‐made polymethyl methacrylate (PMMA) and RANDO phantoms to measure entrance surface air kerma (ESAK). ESAK measurements were also acquired with a 4.5‐cm thick breast phantom for a standard mammogram. Organ dose conversion factors (CFD) were determined as ratio of air kerma at a specific depth to that at the surface for the PMMA phantom and multiplied by the ratio of mass energy absorption coefficients of tissue to air. Normalized eye lens and thyroid gland doses were calculated using the RANDO phantom by multiplying CFD and ESAK values. Maximum variability in OSLD response to scatter radiation from the DBT system was 33% in the W/Rh spectrum and variations in scattered dose distribution were observed between CM and TM. The CFD values for eye lens and thyroid gland ranged between 0.58 to 0.66 and 0.29 to 0.33, respectively. Mean organ doses for two‐view unilateral imaging were 0.24 (CM) and 0.18 (TM) μGy/mAs for the eye lens and 0.24 (CM) and 0.25 (TM) μGy/mAs for the thyroid gland. Higher organ doses were observed during TM compared to CM as the automatic exposure control (AEC) system resulted in greater total mAs values in TM.


| INTRODUCTION
Digital breast tomosynthesis (DBT) is a 3D imaging system that tends to be predominantly used as a diagnostic mammogram for clinical symptoms and as an alternative tool for breast cancer screening. 1,2 The advantages of DBT have been confirmed by several studies [1][2][3][4] and include superior early detection of small cancer, lower recall rate, and improved visualization of breast abnormalities. However, radiation risk from mammography is a significant concern among patients, especially among those undergoing routine breast screening, and their physicians. 5 Therefore, one study by Chetlen et al. 6 quantified exposure in five organs of interest during a routine digital mammography (DM) procedure and disseminated this information to health care providers. They evaluated scatter doses at the skin surface overlaying these five organs in 207 patients using optically stimulated luminescence dosimeters (OSLDs), and found that the mean doses at the sternum, the thyroid gland, the salivary gland, the eye lens, and the uterus were 870, 245, 200, 25, and 11 μGy, respectively. Recently, another study used Monte-Carlo simulations on organ doses from a specific DBT system and reports an increase of up to 21% in thyroid gland and 9% in lung (ipsilateral) during DBT acquisition compared to DM acquisition. 7 However, those organ doses were quantified only from the craniocaudal (CC) view and highly radiosensitive organs such as the eye lens were not considered.
The DBT system provides dual acquisition modes based on DM or DBT acquisitions as there are geometric differences in the acquisition setup between these two modes. Specifically, while the x-ray tube rotates across a compressed breast within a limited angle range during DBT acquisition, it remains fixed during DM acquisition, 8 and the absorbed dose at each acquisition mode also varies when used with the automatic exposure control (AEC) system. 7,9 These differences should be taken into account for estimating scattered dose at organs of interest during DBT imaging. Currently, OSLDs play an importance role in point dose measurement during both radiotherapy and diagnostic imaging due to their characteristics such as high sensitivity, small size, tissue equivalence density, and reusability. While McKeever et al. 10 have reported that OSLDs are capable of measuring doses as low as 10 μGy, linear response among commercially available OSLDs starts at approximately 50 μGy. In addition, previous studies 11,12 have demonstrated the feasibility of using commercial OSLDs in the mammography energy range by adding a correction factor.
Thus, this study aimed to measure and compare scatter doses between DM and DBT acquisitions from a specific DBT system and to estimate absorbed doses in the eye lens and the thyroid gland during two-view mammography under both DM and DBT modes.
We provide data on scattered doses and organ doses and discuss variations in these doses and their potential radiation risk during DM and DBT acquisitions in clinical scenarios.

| MATERIALS AND METHODS
A DBT system (MAMMOMAT Inspiration; Siemens Medical Solutions Inc., Erlangen, Germany) with dual acquisition modes for DM and DBT was used. DM acquisition was used in the conventional mode (CM) while the DBT acquisition was specified as Tomo mode (TM) which acquires 25 projections of breast tissue from different tube angles between −25°and +25°at 2°intervals.

2.A | Radiation dosimeter and correction
The nanoDot OSLD (Landauer Inc., Glenwood, IL, USA) is a small disk made from carbon-doped aluminum oxide (Al 2 O 3 :C) that is enclosed within a light tight plastic frame and measures 10 × 10 × 2 mm. It was used in combination with the microStar reader (Landauer Inc.) for all dose measurements. The lower limit of detection of this OSLD system is 46.7 μGy, according to the calibration certificate provided by the manufacturer. NanoDots were corrected using a parallel-plate ionization chamber (Radcal Corp., Monrovia, CA, USA) for use under specific conditions encountered in our DBT system while measuring scatter radiation generated during mammography. Figure 1 illustrates the experimental setup for nanoDot correction. A 4.5 cm thick breast phantom (CIRS Inc., Norfolk, VA, USA) with an average glandular tissue composition of 50% was positioned for CC view at the center of the image detector with a compression paddle. Five nanoDots were pasted using thin plastic tape and then attached to a pole such that they were similarly placed in air. The ionization chamber was attached to a tripod, and the nanoDots and the ionization chamber were placed symmetrically and laterally at the side of chest wall edge, 4 cm from the center of image detector and 10 cm away from the chest wall edge, such that they were positioned mid-level to the breast phantom (Fig. 1).
The exposure parameters were manually set for the tungsten target and the rhodium filter combinations (W/Rh) with tube voltages of 27, 28, and 29 kV and a tube current-time product of 160 mAs. The breast phantom was irradiated six times at each exposure setting under CM operation and the scatter dose was allowed to accumulate in the nano-Dots for the six irradiations. The nanoDots and the ionization chamber simultaneously recorded the amount of scatter radiation generated predominantly from the irradiated breast phantom. NanoDot correction was repeated twice using a new set of five nanoDots, and finally, all exposed nanoDots were read consecutively three times along with three control nanoDots in the microStar reader to reduce measurement uncertainty; the average of the three readings from each nanoDot was used for calculating the correction factor of the nanoDot (CF OSLD ) as where K IC is average air kerma from the ionization chamber (mGy), K OSLD is average air kerma readout from the five nanoDots (mGy), and BG is average background readout from the control nanoDots (mGy).  The exposure parameters were set for a W/Rh combination of 28 kV and automatic tube current-time product, and this setting was chosen for ESAK measurements because, according to manufacturer's specification, only the W/Rh combination could be modified in TM mode and also because the AEC system identified a tube voltage of 28 kV as optimal for the 4.5 cm compressed breast phantom.
All ESAK measurements were performed in both CM and TM by nanoDots and the nanoDots were irradiated thrice during each measurement to increase the amount of scattered dose received. All The experimental setup for correcting the nanoDots with an ionization chamber for use under conditions encountered during scatter dose measurement.

(a) (b) (c)
F I G . 2. NanoDot positions onto custom-made PMMA phantom at various locations to assess differences in scatter dose received according to angles and distances (a). Scatter radiation dose could be detected from angles of −90°to +90°and at distance ranging from 26.5 to 38.5 cm, depending on the experimental setting; CC view acquisition (b), MLO view acquisition (c).
nanoDots were read thrice along with the three control nanoDots (for measurement of background radiation). The actual scatter dose, in terms of ESAK, was obtained by subtracting background radiation, dividing them by three, and then correcting them based on the CF OSLD value for the W/Rh 28 kV spectrum.

2.C | Organ dose conversion factors
The organ dose conversion factor (CF D ) for the W/Rh 28 kV spectrum was calculated to estimate absorbed doses at organs that received scatter radiation during the mammogram under both CM and TM acquisitions. The nanoDots were inserted into left-and right-sided holes at specific depths in the PMMA phantom (Fig. 3); specifically, two nanoDots each at a depth of 3 mm and two nano-Dots at a depth of 10 mm. Other two nanoDots were pasted onto corresponding locations on the surface of the PMMA phantom.
The positions of the breast and PMMA phantoms and the exposure parameters were identical to ESAK measurements (as described in Section 2.B). Air kerma at the specified depths and the surface were measured separately for the CC and the MLO view acquisitions and in both CM and TM. The air kerma from all exposed nanoDots was measured by reading them thrice and using background subtraction. The CF D was defined as the ratio of air kerma at corresponding depths to surface, and the ratio of mass energy absorption coefficients of tissue to air, and was quantified using the following expression where K depth is average air kerma (mGy) at a specific depth, K surface is average air kerma (mGy) at the surface, μ en ρ tissue is mass energy absorption coefficient of the eye lens and soft tissue (cm 2 /g), and μ en ρ air is mass energy absorption coefficient of air (cm 2 /g). The mass energy absorption coefficients were obtained from the National Institute of Standard and Technology (NIST), 13 according to the effective energy of primary beam, which was estimated using the half value layer of the W/Rh 28 kV spectrum.

2.D | Organ dose estimation
The absorbed doses at the eye lens and the thyroid gland were estimated using the RANDO phantom. A set of 13 nanoDots were pasted on skin of a female RANDO phantom (Phantom Laboratory, Salem, NY, USA) with 10 nanoDots around the eyes and three nano-Dots around the thyroid gland, as shown in Fig. 4(a). Three other nanoDots were used as controls for measuring background radiation.
The RANDO phantom and the breast phantom were positioned simi- The x-ray tube gantry was set at 45°projection in the MLO view.
The cranial RANDO phantom was rotated toward the contralateral side for both CC and MLO views.
Entrance surface air kerma measurements were acquired using four sets of nanoDots for the CC and the MLO views in CM and TM, with a tube voltage of 28 kV and W/Rh combination and automatic tube current-time product settings that were identical to those used for previous ESAK measurements. Three irradiations were conducted for each measurement to allow sufficient accumulation of radiation in the nanoDots, and average of three readout values from each nanoDot, after background subtraction, was divided by three.
The ESAK obtained using the RANDO phantom was used to estimate eye lens and thyroid gland doses using the following equation: where ESAK is the average air kerma readout from the nanoDots, CF OSLD is the correction factor of the nanoDots, and CF D is the organ dose conversion factor.

3.A | NanoDot correction
Scattered radiation sensed by the nanoDots was different from those of the reference dosimeter by 11.9%-32.5%. The reproducibility of scattered dose measurement using nanoDots had a coefficient of variation (CV) of less than 8.7%, as estimated from two measurements. The cumulative scattered dose from six irradiations, CF OSLD , and CV values are listed in Table 1. The CF OSLD at 28 kV was applied for all readout values obtained from the nanoDots.

3.B | ESAK measurements
All nanoDot cumulative doses of less than 50 μGy after three irradiations were excluded from analysis in this study. The measured ESAK values in the PMMA phantom are provided in Table 2 and

3.D | Organ doses
In the RANDO phantom study, negligible radiation doses were recorded at the eye lens in the contralateral imaging side during both CM and TM. The absorbed doses at the thyroid gland and eye lens on the imaging side are shown in     projections at an angular range between −24°and +24°during DBT.
Additionally, data on either breast thickness or compression of the phantom were absent in these reports. These details are crucial as, if the phantom breast was not compressed, it may generate more scatter radiation from the breast and result in higher absorbed dose in the peripheral organs. However, they report an increasing trend in thyroid gland dose during tomosynthesis acquisition which is consistent with the results from our study. Sechopoulos et al. 16  Our results definitively show that exposure of the eye lens and the thyroid gland cannot be avoided during mammography, either in CM or TM, despite the presence of a face shield. Nevertheless, these doses were in the μGy range which is extremely small and seems negligible when compared to the fact that the current threshold dose for a significant risk to organs such as the lens is 0.5 Gy, as suggested by the International Commission on Radiological Protection (ICRP). 18 Further, the thyroid gland is considered less radiosensitive compared to the eye lens.
The above notwithstanding, there are limitations to this study.
We used a specific breast phantom with breast thickness of 4.5 cm and breast composition of 50% glandular tissue and 50% adipose tissue, and a defined x-ray spectrum, which is not completely compatible with the clinical scenario where women who undergo mammography have various breast thicknesses and densities. These factors lead to the use of a variety of x-ray spectra for mammography. For instance, a higher energy spectrum is usually used for thick breasts than that used for the average breast; this may lead to an increase in scattered dose at peripheral organs. Moreover, uncertainty in our results may have occurred due to the angular dependence of nanoDots as we report about 5% difference in scatter radiation compared to Monte-Carlo simulation-based measurements by Okazaki et al. 14

| CONCLUSION
Entrance surface air kerma measurements using OSLDs in a speci-