Journal of Radiological Protection

PAPER Related content

- Comparison of computed tomography
Radiation shielding calculation for digital breast shielding design methods using RadShield
Matthew C DeLorenzo, Kai Yang, Dee H
tomosynthesis rooms with an updated workload Wu et al.

- Average glandular dose in digital

survey mammography and digital breast
tomosynthesis: comparison of phantom
and patient data
To cite this article: Kai Yang et al 2017 J. Radiol. Prot. 37 230 R W Bouwman, R E van Engen, K C
Young et al.

- A Monte Carlo study of the energy spectra

and transmission characteristics of
scattered radiation from x-ray computed
View the article online for updates and enhancements. tomography
David John Platten

This content was downloaded from IP address 128.111.121.42 on 13/03/2018 at 06:49

 | Society for Radiological Protection Journal of Radiological Protection
J. Radiol. Prot. 37 (2017) 230–246 (17pp) https://doi.org/10.1088/1361-6498/aa58a8

Radiation shielding calculation for digital

breast tomosynthesis rooms with an
updated workload survey
Kai Yang1, Thomas J Schultz2, Xinhua Li1 and Bob Liu1
1
Division of Diagnostic Imaging Physics, Department of Radiology, Massachusetts
General Hospital, 55 Fruit Street, Boston, MA 02114, United States
2
Department of Radiology, Massachusetts General Hospital, 55 Fruit Street, Boston,
MA 02114, United States

E-mail: kyang11@mgh.harvard.edu

Received 26 September 2016, revised 20 December 2016

Accepted for publication 11 January 2017
Published 31 January 2017

Abstract
Purpose. To present shielding calculations for clinical digital breast tomo-
synthesis (DBT) rooms with updated workload data from a comprehensive
survey and to provide reference shielding data for DBT rooms. Methods. The
workload survey was performed from eight clinical DBT (Hologic Selenia
Dimensions) rooms at Massachusetts General Hospital (MGH) for the time
period between 10/1/2014 and 10/1/2015. Radiation output related infor-
mation tags from the DICOM header, including mAs, kVp, beam filter mat-
erial and gantry angle, were extracted from a total of 310 421 clinical DBT
acquisitions from the PACS database. DBT workload distributions were
determined from the survey data. In combination with previously measured
scatter fraction data, unshielded scatter air kerma for each room was calcu-
lated. Experiment measurements with a linear-array detector were also per-
formed on representative locations for verification. Necessary shielding
material and thickness were determined for all barriers. For the general pur-
pose of DBT room shielding, a set of workload-distribution-specific trans-
mission data and unshielded scatter air kerma values were calculated using the
updated workload distribution. Results. The workload distribution for Hologic
DBT systems could be simplified by five different kVp/filter combinations for
shielding purpose. The survey data showed the predominance of 45° gantry
location for medial-lateral-oblique views at MGH. When taking into con-
sideration the non-isotropic scatter fraction distribution together with the
gantry angle distribution, accurate and conservative estimate of the unshielded
scatter air kerma levels were determined for all eight DBT rooms. Additional
shielding was shown to be necessary for two 4.5 cm wood doors. Conclusions.

0952-4746/17/010230+17$33.00 © 2017 IOP Publishing Ltd Printed in the UK 230

J. Radiol. Prot. 37 (2017) 230 K Yang et al

This study provided a detailed workload survey and updated transmission data
and unshielded scatter air kerma values for Hologic DBT rooms. Example
shielding calculations were presented for clinical DBT rooms.

Keywords: digital breast tomosynthesis, mammography, workload, transmis-

sion, radiation protection, radiation shielding

(Some figures may appear in colour only in the online journal)

1. Introduction

The current standard for structural shielding design for medical x-ray imaging in the United
States is the National Council on Radiation Protection and Measurement (NCRP), Report No.
147 [1] published in 2004. As the updated version of previous NCRP reports (NCRP Report
No. 34 [2] and NCRP Report No. 49 [3]), the major change of NCRP Report No. 147
included the confined scope in diagnostic x-ray and the decrease of the shielding design goal,
P, which is the permitted radiation level in the occupied area. NCRP Report 147 recom-
mended P=0.1 mGy/week and 0.02 mGy/week for controlled and uncontrolled area,
respectively. The previous corresponding limits in NCRP 49 were 1 and 0.1 mGy/week. As
stated in the NCRP Report No. 147 [1], if only secondary radiation (including scatter and
leakage, while dominated by scatter) shall be considered for breast imaging rooms, three key
components are required to perform accurate shielding calculation: (1) the workload dis-
tribution of a room; (2) the measured scatter fraction data for clinical acquisitions; (3) the
workload-distribution-specific transmission data for different shielding materials determined
from part 1. Simpkin performed scatter intensity measurement, workload survey, and trans-
mission calculations for mammography and summarized the results in two previous studies
[4, 5]. Based on those data, the general recommendation from NCRP Report 147 is that the
typical gypsum boards will be sufficient for mammography rooms. With the wide acceptance
of digital breast tomosynthesis (DBT) systems as the new standard of care for breast cancer
imaging, re-evaluation of necessary radiation shielding for DBT rooms became a critical
medical physics task that is related to the safety of general public and hospital staff. Li et al
first reported the workload distribution survey from 343 DBT exams and calculated trans-
mission data for Hologic Selenia Dimensions (Hologic, Bedford, MA) in 2012 and 2013
[6, 7]. Yang et al [8] measured the scatter fraction data of DBT systems from the same vendor
with a solid-state linear array detector. From that study, a very non-isotropic distribution of
scatter intensities around Holgoic DBT systems was observed, including a relatively high
scatter peak at 25° away from the central x-ray beam. This similar non-isotropic distribution
was observed previously by Judge et al [9] from the 2D mammographic version by the same
vendor (Hologic Selenia). This finding emphasizes the importance of the gantry’s angular
distribution when calculating the shielding requirement. Furthermore, the imaging technique
such as the contrast to noise ratio correction factors used in AEC has been optimized by the
manufacturer since the survey by Li et al [7]. This becomes the major motivation to perform
an updated workload survey for DBT systems with special attention paid to the gantry angle
distribution.
In this work, we performed a comprehensive survey from eight DBT rooms at Massa-
chusetts General Hospital (MGH) for a period of 1 year. Unshielded scatter air kerma were
calculated using previously published scatter fraction data [8], in combination with the
workload distribution from the new survey. The goal was to verify the adequacy of the current

231
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Table 1. DICOM tag list from Hologic DBT images.

DICOM
TAG Field name
0008 0020 Study date
0008 1010 Station name
0008 1030 Study_description
0008 103e Series_description
0008 0060 kVp
0018 1030 Protocol_name
0018 1150 Exposure time
0018 1151 X_ray_tube_current
0018 1152 Exposure (in the unit of mAs)
0018 1153 Exposure_in_μAs
0018 1190 Focal spots
0018 1191 Anode_target_material
0018 11a0 Body part thickness
0018 11a4 Paddle_description
0018 1405 Relative_X_ray_exposure
0018 1510 Positioner_primary_angle
0018 5101 View_position
0018 7050 Filter_material
0018 7060 Exposure_control_mode
0018 7062 Exposure_control_mode_description
0018 8150 Exposure_time_in_us
0020 0060 Laterality
0020 1002 Images_in_acquisition
0040 0314 HVL
0040 0316 Organ_dose
0040 8302 Entrance_dose_in_mGy

shielding for DBT rooms in our hospital and to provide updated shielding information for the
general field of medical physics. For the latter purpose, the unshielded scatter air kerma at 1 m
and workload-distribution-specific transmission data determined based on the previous
foundation work by our group [6–8] are the end results of this study.

2. Methods and materials

2.1. DBT workload survey

A total of eight DBT rooms (all with Hologic Selenia Dimensions) at MGH main campus
were included in this workload survey. Among them, four rooms are designated mainly as
screening rooms and four rooms are mainly designated as diagnostic rooms. For the time
period between 10/1/2014 and 10/1/2015, 310 421 clinical exposures (2D or 3D) for 36 145
patients from these rooms were recorded and stored as DICOM images in our PACS system.
Key DICOM header tags from each image were automatically extracted (table 1). The
extracted information were congregated into Microsoft Excel files and analyzed by the sta-
tistics software package, R (version 3.2.4) [10].
The workload distribution of each room was summarized by the mAs of each kVp within
the entire clinical kVp range for each acquisition mode (2D versus 3D). A simplified

232
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Figure 1. Unshielded scatter air kerma calculation models. Top row: ceiling; middle
row: door/wall; bottom row: floor. The shaded fan region indicates the solid angle
region where scatter radiation was considered for a barrier. The angular locations of
maximal unshielded air kerma marked by bold arrows were also indicated with text.

workload distribution was further derived with 5 kV bins while maintaining conservatism in
the followed calculation steps. The gantry’s angular distribution for each acquisition view was
also analyzed by the mAs weighting. The views typically include craniocaudal (CC), medial-
lateral-oblique (MLO), medial-lateral (ML), and lateral-medial (LM) etc.

2.2. Unshielded scatter air kerma calculation

Following a previously developed approach [8], unshielded scatter air kerma for each DBT
room was calculated by including the updated workload distribution using the most repre-
sentative filter/kVp combination determined from the workload survey. For each filter/kVp
combination (inherently associated with the 2D/3D acquisition mode for Hologics DBT
systems), the unshielded scatter air kerma was calculated by considering the gantry position
associated with different clinical views and the angular distribution of the scatter fraction data
[8]. The results were summed for all the acquisition view as:
Ksca (i , k ) = åW (i, j) KP (i, j) a max (i, j) f ( j, k ), ( 1)
j

where Ksca (i, k ) is the unshielded scatter air kerma for a specific combination of barrier
(indexed by k) and kVp/filter group (indexed by i). W (i, j) is the workload (mAs) for a

233
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Table 2. Shielding related parameters for two representative rooms.

Room ID Screening room BI7 Diagnostic room BI2

Area description (Left) Corridor door Corridor door

Distance to phantom center (m) 2.3 2.7
Occupancy factor, T 0.125 0.125
Shielding goal, P (mGy wk−1) 0.02 0.02
Area Description (Right) Waiting room Corridor
Distance to phantom center (m) 1.98 2.7
Occupancy factor, T 0.05 0.2
Shielding goal, P (mGy wk−1) 0.02 0.02

specific kVp/filter group (indexed by i) and acquisition view (indexed by j, e.g. CC, MLO,
ML, etc). KP (i, j) is the primary output air kerma per unit workload (mGy/mAs) measured at
1 m from the focal spot with the same index as W (i, j) and the values were previously
reported [8]. a max (i, j) is the corresponding scatter fraction (cm−1) that provides the maximal
scatter air kerma to the barrier (as shown in marked locations in figure 1). f ( j, k ) is a scalar
(with unit of cm) which takes into consideration of the primary beam area and the DBT gantry
center to barrier distance. For this study involving Hologic DBT systems, the primary beam
area was assumed to be fixed at 24 × 30 cm2 for simplicity and conservatism, as previously
discussed [8].
Figure 1 shows an example of the basic geometry related to the calculation. For a
standard room setup, the DBT gantry center is 1.2 m above the floor, 2 m away from adjacent
wall/door. The ceiling is 3.05 m above the floor. The shielding for the wall/door is only
considered up to 2.13 m from the floor. The maximal scatter air kerma was identified for a
specific barrier based on the solid angle range it is facing. Although the scatter air kerma for
both the ceiling and floor were calculated, the focus was on the wall/door and thus only the
wall/door results are reported in this study for brevity. For each room, the left and right
direction was defined when facing the front of the DBT system.
We would like to emphasize that the calculation approach described in this section and
the next section can be generally applicable to any DBT rooms, as long as the angular
distribution data of scatter fraction and workload are available.

2.3. Barrier thickness calculation

For each room, the related barrier distance, shielding goal, and occupancy factor of adjacent
area were determined per the recommendation of NCRP Report No. 147 [1]. Table 2 sum-
marizes the related parameters for two example DBT rooms (BI7 for screening and BI2 for
diagnostic).
Li et al [6, 7] published primary transmission data of Hologic DBT systems for all three
beam filter materials (Rh, Ag, and Al) with a wide kVp range and the data were summarized
by the corresponding α, β, γ fitting parameters used in the Archer equation [1]. Under the
conservative assumption that the scattered radiation should have similar transmission property
as the primary radiation, the transmission data from Li et al [6, 7] was directly used for each
individual kVp/filter combination to determine the total barrier thickness in this study. For a
specific barrier indexed by k and the corresponding shielding goal of P(k) and occupancy
factor T(k), the proper barrier thickness was iteratively determined by satisfying the below

234
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Figure 2. Setup for experimental validation. (a): The experiment diagram. The linear
detector was placed at three different vertical positions outside the wood door. (b):
Photos of the experiment setup. Notice the thin piece of metal at the bottom of the door.

condition:
P (k )
å i Ksca (i, k ) B (i, x)  T (k )
, ( 2)

where again i is the index for kVp/filter combination and for each individual i, the
transmission B can be determined by the Archer equation as:

- 1
⎡⎛ b⎞ b⎤ g
B (x ) = ⎢ ⎜1 + ⎟ eagx - ⎥ . (3)
⎣⎝ a⎠ a⎦
The Archer equation [11] is an empirical model to describe the transmission through a
barrier with certain thickness of x mm. The three parameters (α, β and γ) depend on the
workload distribution and the barrier material.

235
J. Radiol. Prot. 37 (2017) 230 K Yang et al

2.4. Validation through experiment

To validate the above calculation of unshielded scatter air kerma and related barrier trans-
mission, a spot check experiment was performed for one DBT room (BI7) at MGH. With a
solid state linear array detector (X-Scan 0.8f3-512, Detection Technology Inc., Finland), the
scatter air kerma was measured directly outside of the room entrance (2.3 m away from the
gantry center, as shown in figure 2), with and without the wood door closed. The gantry was
positioned at 45° for a typical MLO acquisition. At three different vertical locations (shown in
figure 2), two different DBT exposures were performed for a 6 × 10 × 13 cm3 acrylic
phantom: 2D 35 kVp/Ag 400 mAs and 3D 40 kVp/Al 100 mAs, again with and without the
wood door closed. At the detector position 1 and 2, additional 3D exposures were also
performed for 30 kVp/Al 100 mAs and 35 kVp/Al 100 mAs. A previously developed cali-
bration process [8] was applied to convert the measured ADU values into air kerma levels.
The measured scatter air kerma and related transmission through a 4.5 cm thick wood door
was compared to the calculation results from sections 2.2 to 2.3.

2.5. Workload-distribution-specific transmission data for typical shielding materials and

unshielded scatter air kerma at 1 m

Although it is more accurate to calculate the barrier thickness with equations (1) and (2) by
considering the transmission data for each individual kVp/filter combination, a single set of
generic/representative transmission data and unshielded air kerma at 1 m would be useful
from a practical perspective. Many examples are shown in the NCRP Report No. 147 [1]
including the workload-distribution-specific transmission. Generally the barrier thickness can
be estimated without knowing the exact workload distribution of a certain room. For that
purpose, we calculated a set of generic transmission data for scatter from Hologic DBT
systems, using the results from Li et al [6, 7] and the updated workload distribution from this
study. The workload-distribution-specific transmission, Bw (x ), through a barrier with certain
thickness, x, can be determined as:
Bw ( x ) = å i Ksca (i) B (i, x), ( 4)
where i is the index for each kVp/filter combination from this workload survey, Ksca (i ) was
Ksca (i, k ) in equation (1), with the barrier index k dropped and B (i, x ) was determined
through equation (3).
The corresponding Archer equation (3) fitting parameters were determined by curve
fitting on Bw (x ) and were reported as the workload-distribution-specific transmission data for
scatter radiation from a representative Hologic DBT room.
For the unshielded air kerma at 1 m, a simple but conservative estimate should be
provided if the exact workload for a specific room is unknown. Given the non-isotropic
distribution of scatter radiation, the decision was made to provide three sets of unshielded air
kerma per view at 1 m as:
1 1
1. Ksec _ CC _ 2D and Ksec _ CC _ 3D, for the CC view acquisitions, under 2D and 3D mode
respectively;
_ T _ 2D and Ksec _ T _ 3D, for the oblique view acquisitions (including MLO, ML, etc)
1 1
2. Ksec
with the primary beam ‘toward’ the barrier;
_ A _ 2D and Ksec _ A _ 3D, for the oblique view acquisitions (including MLO, ML, etc)
1 1
3. Ksec
with the primary beam ‘away’ from the barrier.

236
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Table 3. Parameters used for calculation of unshielded scatter air kerma at 1 m. CC:
cranial caudal; T: oblique toward; A: oblique away. Scatter angle was defined as
between 0° and 180°, with 0° toward the DBT detector and 180° toward the x-ray tube.
View_mode Max scatter angle (degree) mAs/view

CC_2D 164.5 135.0

T_2D 25.0 144.3
A_2D 164.5 144.3
CC_3D 158.0 61.9
T_3D 28.0 62.9
A_3D 158.0 62.9

Table 3 summarizes the parameters used to calculate the unshielded air kerma at 1 m.
Basically for the CC view and oblique view with primary beam away from the barrier, the
maximal back scatter air kerma around 160° was used, just exactly the same assumption as
the NCRP Report No. 147. For the oblique view with primary beam toward the barrier, the
high scatter peak around 25° (28° for 3D mode) was used, with a conservative assumption
that this peak will project onto the barrier. The kV/filter distribution for each view/mode
combination was determined from the survey performed in this study (table 5).
With these data, a user can follow the methodology of NCRP Report No. 147 and
directly calculate the necessary shielding for a representative DBT room. The other para-
meters needed are: barrier distance (d in meter) and workload in views/week for CC views
(NCC _ 2D and NCC _ 3D ) and oblique views (NT _ 2D, NT _ 3D, NA _ 2D, and NA _ 3D ). The total
unshielded air kerma (in the unit of mGy/week) at a certain barrier distance can be calculated
as:

1
Ksec =
å i Ksec
1
_ i Ni
, (5)
d2

where i is the index for six different view/mode combinations just described above.

3. Results

3.1. Survey results

Table 4 shows the basic information from the workload survey. The screening rooms were
much busier with ∼114 patients/week and 8.98 views/patient as compared to ∼61 patients/
week and 7.80 views/patient for the diagnostic rooms. The screening rooms had a 50/50 split
between CC and MLO views, while the diagnostic rooms have about 25% of ML/LM views
and a lower percentage of MLO views. It is also interesting to observe that the mAs/view is
always higher for the MLO view than the CC view, due to the inclusion of chest wall muscle
in the MLO view. The mAs/view values in the four highlighted boxes were used to calculate
the typical unshielded scatter air kerma in section 3.4.
Figure 3 shows the DBT gantry angle distribution. The gantry angle locations are
straightforward for the CC and ML/LM views, at 0° and ±90°, respectively, and are pro-
portional to their frequencies of usage (as shown by the mAs/view and views/patient data in
table 1). For the gantry angle distribution of MLO views, it really depends on patient con-
dition and local imaging practice. The data in figure 3 shows that at MGH, the MLO views
were very concentrated at gantry angles of ±45°. Therefore in the following shielding

237
 J. Radiol. Prot. 37 (2017) 230
Table 4. Basic work load information.

Screening rooms Diagnostic rooms

Room ID BI1 BI3 BI6 BI7 Screening rooms average BI2 BI4 BI5 BI10 Diagnostic rooms average

Total patients 6789 5820 5566 5388 5891 1792 3363 4057 3370 3146
Patients/week 131 112 107 104 114 34 65 78 65 61
Total views 59 003 56 180 48 627 47 694 52 876 13 532 25 651 33 825 25 909 24 729
Views/patient 8.69 9.65 8.74 8.85 8.98 7.55 7.63 8.34 7.69 7.80
Total 2D views 32 457 27 626 27 062 26 603 28 437 9158 15 026 19 236 15 001 14 605
2D views/patient 4.78 4.75 4.86 4.94 4.83 5.11 4.47 4.74 4.45 4.69
2D mAs/view 138 137 140 137 137.8 146 139 129 131 136.1
2D CC views/patient 2.43 2.4 2.5 2.54 2.47 2.43 2.13 2.41 2.11 2.27
2D CC mAs/view 136 135 137 135 135.5 139 134 127 127 131.7
238

2D MLO views/patient 2.35 2.35 2.36 2.4 2.37 1.02 1.43 2.12 1.46 1.51
2D MLO mAs/view 140 139 143 139 140.2 157 150 131 138 144.3
2D ML views/patient 1.65 0.9 0.22 0.89 0.92
2D ML mAs/view 148 133 133 127 135.0
Total 3D views 26 546 28 554 21 565 21 091 24 439 4374 10 625 14 589 10 908 10 124
3D views/patient 3.91 4.91 3.87 3.91 4.15 2.44 3.16 3.6 3.24 3.11
3D CC views/patient 1.96 2.45 1.94 1.96 2.08 1.08 1.4 1.76 1.44 1.42
3D mAs/view 63 60 56 69 62.4 61 63 53 58 58.8
3D CC mAs/view 63 60 56 69 61.9 61 63 52 57 58.5
3D MLO views/patient 1.96 2.45 1.93 1.95 2.07 0.88 1.25 1.72 1.29 1.29
3D MLO mAs/view 64 61 57 70 62.9 62 64 53 59 59.3
3D ML views/patient 0.49 0.51 0.11 0.51 0.41
3D ML mAs/view 60 63 53 57 58.0

K Yang et al
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Figure 3. Gantry angle distribution weighted by mAs. The high peaks around 0°
correspond to the CC views. The peaks around ±45° correspond to the MLO views.
The peaks around ±90° correspond to the ML/LM views.

Figure 4. kVp distributions as weighted by total mAs. (a) 2D mode; (b) 3D mode. Data
were shown for integer kVp values from the DICOM header. Average data from all
eight rooms are shown as solid lines.

calculation, the MLO views are assumed to be equally distributed between +45° and −45°
gantry angle (as shown in the central panel of figure 1).
Figure 4 shows the workload distributions (weighted by the total mAs) for the 2D and 3D
mode. There was not a major difference between screening rooms and diagnostics rooms.
However, the 2D and 3D mode had very different workload distributions.
For the 2D mode, only 0.30% of total workload was from kVp 36 and for the 3D
mode, only 1.48% from kVp 41. Therefore to simplify the calculation for scatter air kerma
and transmission, the above distributions were re-binned into five major groups (table 5
below) while still maintaining conservatism for shielding purpose.
It is worth noting that for Hologic DBT systems, two different filters can be used for the
2D mode: rhodium (Rh) and silver (Ag). As previously shown, for the same kVp and imaging
mode, the filter selection had little impact on the scatter fraction [8]. Therefore in this study,
the calculation used for the 2D mode was based on 30 and 35 kVp/Ag for simplicity. For 3D
mode, 30, 35 and 40 kVp/Al were used. Scatter fraction data were available from our

239
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Figure 5. Scatter air kerma measurements with the door open and closed. (a) 3D mode
with 40 kVp/Al; (b) 2D mode with 35 kVp/Ag. The scatter signal was measured with
100 mAs for the primary technique and from three consecutive vertical positions of the
linear detector (as three bands for each group of data).

Table 5. Workload distribution (percentage in mAs) used for scatter and transmission
calculation.
2D 3D
kVp Distribution kVp Distribution
30 57.3% 30 23.6%
35 42.7% 35 60.9%
40 15.6%

previous study [8]. The data in table 4 was used in the scatter calculation to distribute the total
workload (mAs for either 2D or 3D mode) for each room.

3.2. Experimental validation

Figure 5 shows the scatter air kerma measured by the linear array detector for two different
kVp/filter combinations: 40 kVp/Al for 3D and 35 kVp/Ag for 2D. The high peak levels of
scatter air kerma confirmed with the 25° peak of the scatter fraction data and the calculated
locations for 45° gantry angle shown in figure 1. The attenuation of the scatter air kerma by
the 4.5 cm wood door was apparent. Using the less noisy data from the range between 0 to
50 cm, the transmission data from the wood door can be calcualted and is shown in figure 6.
An average of tranmission of 0.28 and 0.35 can be estimated for 35 kVp/Ag and 40 kVp/Al,
respectively. Using the data from Li et al [6, 7], the transmissions for 45 mm wood for the
same sets of kVp/filter combination are 0.341 and 0.422, respectively. Although closely
matching, the slightly lower transmission from the measurement was likely due to the oblique
incident angle and the additional piece of thin metal plate on the door (figure 2(b)).
Figure 7 shows the scatter air kerma measured for the 3D DBT mode. At the same
primary mAs setup, the average of 30 and 40 kVp data is shown to be a very close estimate of

240
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Figure 6. Transmission data measured through a 4.5 cm wood door.

Figure 7. Scatter air kerma measured with different kVps for the 3D mode. The scatter
signal was measured with 100 mAs for the primary technique. The averaged data
between 30 and 40 kVp (solid line) is shown to be a good estimate of the 35 kVp data
(circular marker).

the 35 kVp data. Thus it validated the assumption we used for the scatter air kerma calculation
below.

3.3. Unshielded scatter air kerma calculation and barrier thickness results

Following the same calculation process from our previous study [8] and with the geometry
assumptions shown in figure 1, unshielded scatter air kerma was calculated for all eight DBT
rooms. The calculations were performed based on the workload and kVp distribution from the
above mentioned survey results. Table 6 shows the results from two representative rooms
(BI7 for screening and BI2 for diagnostic), with the unshielded scatter air kerma, Ksca (i, k ),
for each kVp/filter combination and barrier location (only for the wall/door at the left and
right side).

241
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Table 6. Unshielded scatter air kerma per week (mGy/week) by kVp distribution.

Room ID Screening room BI7 Diagnostic room BI2

Left: door 30 kVp Ag_2D−1 0.53 0.10
35 kVp Ag_2D−1 0.67 0.11
30 kVp Al_3D−1 0.07 0.01
35 kVp/Al_3D 0.32 0.04
40 kVp Al_3D−1 0.11 0.01
Right: wall 30 kVp Ag_2D−1 0.72 0.10
35 kVp Ag_2D−1 0.91 0.11
30 kVp Al_3D−1 0.10 0.01
35 kVp Al_3D−1 0.43 0.04
40 kVp Al_3D−1 0.16 0.01

Table 7. Barrier transmission calculation results. Both rooms have doors on the left and
walls on the right. The 28 mm gypsum boards provided sufficient shielding for both
rooms’ right side since the scatter air kerma transmitted through barrier was below the
shielding goal. Additional shielding (e.g. a 20 gauge sheet of steel) is shown to be
necessary for the door of BI7 to achieve the shielding goal.
Screening Diagnostic
Room ID room BI7 room BI2
Left: door Shielding goal, P/T 0.16 0.16
(mGy wk−1)
Scatter air kerma transmitted 45 mm wood for door 0.5883 0.0958
through barrier (mGy wk−1)
45 mm wood+20 gauge 0.0001 —
(0.953 mm) steel
Right: wall Shielding goal, P/T 0.40 0.10
(mGy wk−1)
Scatter air kerma transmitted 28 mm gypsum for wall 0.0096 0.0010
through barrier (mGy wk−1)

Based on equations (1) and (2), the necessary thickness of shielding materials (gypsum
for wall and wood for door) for each room was calculated using the transmission data from Li
et al [6, 7]. Table 7 summarizes the example results for BI7 and BI2.
Similar to the example results of BI2, most of the rooms have sufficient shielding with
the current available 28 mm gypsum dry wall and 45 mm wood doors, except for two rooms
(BI4 and BI7). The main reason was because the side of gantry was directly facing the door
for these two rooms and thus the 25° scatter peak was projected onto the door for ‘obli-
que_toward’ views. By adding an additional layer of 20 gauge steel (0.953 mm), these two
doors will meet their shielding goals.

3.4. Workload-distribution-specific transmission data and unshielded scatter air kerma at 1 m

for a representative DBT room

Figures 8–10 show the workload-distribution-specific transmission data for typical shielding
materials using the workload distribution from the survey results. Table 8 summarizes the
corresponding fitting parameters for these curves using the Archer equation (including the

242
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Figure 8. Workload-distribution-specific DBT scatter transmission data for lead and

steel.

Figure 9. Workload-distribution-specific DBT scatter transmission data for gypsum,

concrete, and glass.

fitting parameter R2). It is worth noting that these transmission curves represent more
penetrating beam than the data used in the NCRP Report No. 147 [1].
Table 9 provides reference data on unshielded scatter air kerma at 1 m for six repre-
sentative DBT views/mode combinations in the unit of mGy/view.
The reference data provided in tables 8 and 9 were intended to help medical physicists to
perform shielding calculations following the NCRP Report No. 147 methodology, without
going through the complex calculation using the scatter fraction data. For example, if we
assume a DBT room with a workload of 100 patients/week and all the exams are 2D+3D
screening exams for both breasts, then:
NCC _ 2D = NCC _ 3D = 200 views,

NT _ 2D = NT _ 3D = NA_ 2D = NA_ 3D = 100 views.

243
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Figure 10. Workload-distribution-specific DBT scatter transmission data for wood.

Table 8. Archer equation fitting parameters (see equation (3)) for scatter transmission
from a typical Hologic DBT room.
α β γ R2

Lead 17.2400 66.9600 0.2159 0.999 995

Concrete 0.1385 0.6029 0.3138 0.999 994
Steel 3.3840 13.9500 0.2834 0.999 993
Gypsum 0.0629 0.3191 0.2934 0.999 986
Glass 0.1479 0.6545 0.3063 0.999 995
Wood 0.0124 0.0131 0.2146 0.999 999

Table 9. Unshielded scatter air kerma per view (mGy/view) at 1 m.

2D 3D
CC 6.07 × 10−3 7.85 × 10−3
Oblique_toward (T) 4.83 × 10−2 1.43 × 10−2
Oblique_away (A) 6.49 × 10−3 7.89 × 10−3

For a fully occupied office (T = 1) which is d = 2 m away from the gantry, the unshielded
scatter air kerma can be determined with table 9 as:

Ksec (0) =
200 ´ (6.07 ´ 10-3 + 7.85 ´ 10-3) + 100 ´ (4.83 ´ 10-2 + 1.43 ´ 10-2 + 6.49 ´ 10-3 + 7.89 ´ 10-3)
22
= 2.62 mGy /week.

With a shielding goal of P=0.02 mGy/week, the transmission from a certain barrier is:

0.02
B (x ) = = 7.6 ´ 10-3.
2.62
Using the transmission data from table 8, a minimum of 23 mm gypsum is needed.

244
J. Radiol. Prot. 37 (2017) 230 K Yang et al

4. Discussion

Motivated by the findings from our previous DBT scatter measurement [8], in this study we
aimed to perform detailed shielding calculation with an updated workload survey. Due to the
high scatter air kerma peak around 25° for Hologic Selenia Dimensions DBT systems [8], it is
critical to have a workload distribution on gantry angle for each clinical DBT views. The
major concern related to DBT shielding depends on the projected location of this peak onto a
certain barrier. Specifically for MGH, the MLO views are exclusively acquired at gantry
angles of +45° and −45° and therefore a high percentage (∼50%) of DBT acquisitions will
have this high scatter peak projected onto a wall/door, instead of the floor. Besides the
distance to the barrier, the gantry height also affects the vertical location of this peak. To
maintain the traditional conservatism related to radiation shielding for mammography rooms
(in which a uniform distribution with maximum scatter air kerma was assumed [1]), we
accepted this high peak location as our shielding focus, although in our calculation it was only
projected to the barrier at ∼10–20 cm above the floor.
Even with the above conservative inclusion of the high scatter peak onto a wall/door and
busy clinical workload at MGH, the current double layer of gypsum board (28 mm) was
shown to be sufficient for shielding, with the only exception of two wood doors. For these
two cases, the DBT system was at least 2 m away from the barrier. Therefore this suggests a
practical solution for DBT room installation and shielding: the room door should avoid facing
the side of the gantry, otherwise additional shielding material might be necessary for a
standard 4.5 cm wood door, depending on the distance and workload.
It is also interesting to compare the data provide in table 9 to the recommended
unshielded scatter air kerma at 1 m from NCRP Report No. 147, 3.6 ´ 10-2 mGy/patient.
For a typical screening DBT exam for both breasts, it will be 2 CC_2D views, 2 CC_3D
views, 1 oblique_2D_toward, 1 oblique_3D_toward, 1 oblique_2D_away, and 1 oblique_3D
away. By simply adding up these corresponding numbers from table 9, we got
1.05 ´ 10-1 mGy /patient with this workload survey from MGH. The key difference comes
from the added 3D views and the 25° scatter peak at the ‘oblique_toward’ views.
As opposed to traditional 2D mammography, the current available clinical DBT systems
have very different system designs and acquisition schemes among different manufacturers.
We would expect that the unshielded scatter air kerma distribution and transmission data will
all be system dependent [12]. Besides our study on the Hologic Selenia Dimensions system,
there is no other information currently available for other manufacturers. But as recommended
by both NCRP Report No. 147 (for United States) and the 2012 report of the British Institute
of Radiology working party on radiation shielding for diagnostic radiology [13] (for United
Kingdom), a DBT system should be best mounted on the wall opposite the door. In that
layout, the patient will effectively shield the door from the scattered radiation, and no
additional shielding will be required for the door, just as the example in this study indicated.

5. Conclusions

This study provided a detailed workload survey and updated reference data for shielding
calculations for Hologic DBT rooms. Example calculations were presented on eight DBT
rooms in MGH and the accuracy was validated with experimental measurements.

245
J. Radiol. Prot. 37 (2017) 230 K Yang et al

Acknowledgments

The authors would like to acknowledge Tomi Fält from Detection Technology, Inc. for the
support of the linear-array detector and Matthew C DeLorenzo from MGH for constructive
discussions related to this manuscript.

References

[1] NCRP National Council on Radiation Protection and Measurement 2004 Structural shielding
design for medical x-ray imaging facilities NCRP Report Number 147
[2] NCRP National Council on Radiation Protection and Measurement 1970 X-ray and gamma-ray
protection for energies up to 10 MeV—structural shielding design and evaluation NCRP Report
No. 34
[3] NCRP National Council on Radiation Protection and Measurement 1976 Structural shielding
design and evaluation for medical use of x rays and gamma rays of energies up to 10 MeV
NCRP Report No. 49
[4] Simpkin D J 1995 Transmission data for shielding diagnostic x-ray facilities Health Phys. 68
704–9
[5] Simpkin D J 1996 Scatter radiation intensities about mammography units Health Phys. 70 238–44
[6] Li X, Zhang D and Liu B 2012 Transmission of broad W/Rh and W/Al (target/filter) x-ray beams
operated at 25–49 kVp through common shielding materials Med. Phys. 39 4132–8
[7] Li X, Zhang D and Liu B 2013 Workload and transmission data for the installation of a digital
breast tomosynthesis system Med. Phys. 40 063901
[8] Yang K, Li X and Liu B 2016 Scatter radiation intensities around a clinical digital breast
tomosynthesis unit and the impact on radiation shielding considerations Med. Phys. 43
1096–110
[9] Judge M A, Keavey E and Phelan N 2013 Scatter radiation intensities around full-field digital
mammography units Br. J. Radiol. 86 20120130
[10] R Development Core Team 2016 R: A Language and Environment for Statistical Computing
(Vienna: R Foundation for Statistical Computing)
[11] Archer B R, Thornby J I and Bushong S C 1983 Diagnostic x-ray shielding design based on an
empirical model of photon attenuation Health Phys. 44 507–17
[12] Sechopoulos I et al 2014 Radiation dosimetry in digital breast tomosynthesis: report of AAPM
Tomosynthesis Subcommittee Task Group 223,’ Med. Phys. 41 091501
[13] Sutton D G, Martin C J, Williams J R and Peet D J 2012 Radiation Shielding for Diagnostic
Radiology ( London: Charleswrorth)

246