Dosimetric aspects of inverse-planned modulated-arc total-body irradiation

Mareike Held, Neil Kirby, Olivier Morin, and Jean Pouliota)

Department of Radiation Oncology, University of California San Francisco, California 94143-1708
(Received 30 April 2012; revised 9 July 2012; accepted for publication 11 July 2012; published 2
August 2012)
Purpose: To develop optimal beam parameters and to verify the dosimetric aspects of the recently de-
veloped modulated-arc total-body irradiation (MATBI) technique, which delivers an inverse-planned
dose to the entire body using gantry rotation.
Methods: The patient is positioned prone and supine underneath the gantry at about 2 m source-
to-surface distance (SSD). Then, up to 28 beams irradiate the patient from different gantry angles.
Based on full-body computed-tomography (CT) images of the patient, the weight of each beam is
optimized, using inverse planning, to create a uniform body dose. This study investigates how to
best simulate patients and the ideal beam setup parameters, such as field size, number of beams, and
beam geometry, for treatment time and dose homogeneity. In addition, three anthropomorphic water
phantoms were constructed and utilized to verify the accuracy of dose delivery, with both diode array
and ion chamber measurements. Furthermore, to improve the accuracy of the new technique, a beam
model is created specifically for the extended-SSD positioning for MATBI.
Results: Low dose CT scans can be utilized for dose calculations without affecting the accuracy. The
largest field size of 40 × 40 cm2 was found to deliver the most uniform dose in the least amount of
time. Moreover, a higher number of beams improves dose homogeneity. The average dose discrep-
ancy between ion chamber measurements and extended-SSD beam model calculations was 1.2%,
with the largest discrepancy being 3.2%. This average dose discrepancy was 1.4% with the standard
beam model for delivery at isocenter.
Conclusions: The optimum beam setup parameters, regarding dose uniformity and treatment dura-
tion, are laid out for modulated-arc TBI. In addition, the presented dose measurements show that these
treatments can be delivered accurately. These measurements also indicated that a new beam model
did not significantly improve the accuracy of dose calculations. The optimum beam setup parame-
ters along with the measurements performed to ensure accurate dose delivery serve as a useful guide
for the clinical implementation of MATBI. © 2012 American Association of Physicists in Medicine.
[http://dx.doi.org/10.1118/1.4739250]

Key words: total-body irradiation (TBI), treatment techniques, inverse planning, full-body dose
distribution, TBI arc treatment

I. INTRODUCTION supine and prone positions on the treatment couch, which is

Patients receive total-body irradiation (TBI) treatment in
preparation for a hematopoietic stem-cell transplant. The ra-
diation dose during this treatment can be lethal, which is why
precise dose delivery is essential. This can be challenging,
as the target volume is highly irregular and also larger than
the maximum field size at nominal treatment distances. Con-
sequently, to cover the entire target, there are two possibili-
ties. Either the source-to-surface distance (SSD) is extended
so that one large field covers the entire patient1, 2 or multi-
ple fields are utilized.3–5 Incorporating the latter with mod-
ern technology allows for intensity modulation to account for
variation in patient thickness and internal heterogeneities.6
With this in mind, a new CT-based inverse-planned TBI treat-
ment method has been developed for which the patient is
alternated between supine and prone setups on a treatment
couch near the floor at about 2 m SSD.7, 8 The treatment couch
is a modified massage couch, which provides comfortable po-
sitioning throughout the entire treatment. The treatment couch
height is constant, whereas the SSD is patient specific. This is
accounted for by measuring the SSD of the patient in both
then applied in the treatment planning system. In contrast to
other inverse-planned TBI methods, gantry rotation is used
to scan the radiation field across the patient. Depending on
the patient size, between 16 and 28 open field beams per side
are necessary to cover the entire target volume. The weight
for each beam is varied to optimize body-dose uniformity.
Figure 1 shows the setup for supine positioning during
modulated-arc total-body irradiation (MATBI).
The developed technique has many degrees of freedom;
thus, many steps are necessary to ensure optimal treatment.
This study finds the most favorable parameters to implement
MATBI clinically, focusing on how to optimally simulate,
plan, and deliver this TBI technique. There is an investiga-
tion of the ideal CT scanning parameters to use during the
full-body imaging. Different treatment beam parameters are
compared here to demonstrate which are optimal with respect
to dose homogeneity and treatment time. The patient treat-
ment positioning near the floor also requires the analysis of
floor backscatter effects and evaluation of a new beam model
specifically for treatment at extended SSD. Since the patient
is repositioned between supine and prone treatments, there is

5263 Med. Phys. 39 (8), August 2012 0094-2405/2012/39(8)/5263/9/$30.00 © 2012 Am. Assoc. Phys. Med. 5263
5264 Held et al.: Dosimetric aspects of modulated-arc TBI
F IG . 1. Supine patient setup for modulated-arc total body irradiation with N
beams.
a potential for dosimetric errors from misalignment. The ef-
fects of these errors are studied here. Physical blocks that re-
duce the dose to an organ at risk (OAR) are included in the
final dose computation, which requires verification of dose
transmission calculation in the treatment planning software
(TPS). Additionally, there is an investigation into the effect of
rotational delivery across these blocks. Three anthropomor-
phic phantoms were constructed for dose measurements in the
course of this work.
II. MATERIALS AND METHODS
An advantage of MATBI is that it provides access to full-
body dose distributions. This requires a detailed analysis of
all the aspects that might impact the final dose delivered to
the patient. The following aspects have been investigated here
to ensure that these treatments are optimally and accurately
delivered:
r CT scanning parameters.
r Treatment planning beam model.
r Backscatter off the floor.
r Treatment beam parameters.
r Beam setup errors.
r Dose delivery accuracy.
r Modeling blocks for organ shielding.
r Rotational delivery across blocks.
II.A. Optimal CT scanning parameters
A full-body CT scan is acquired of each patient in the
supine position for treatment planning. To reduce the amount
of data and the time required for contouring, images are taken
with lower quality settings than standard treatment scans.
Lowering the image quality also reduces the dose exposure of
the patient during the scan. To test the effect of reduced im-
age quality on the dosimetry, a Rando phantom was scanned
Medical Physics, Vol. 39, No. 8, August 2012
5264
with different CT parameters, i.e., image slice thickness, tube
current, and peak kilo-voltage. The same radiation plan was
calculated to the different CT scans and the CT dose in-
dex (CTDI), or the average absorbed dose per slice by a
patient,9, 10 was evaluated based on the dosimetric accuracy.
CT scans were acquired with a CTDI of 1.4 and 20.0 mGy.
II.B. Treatment planning system
All treatment plans are created using the Pinnacle 9.0 TPS.
As suggested by Lavallée et al.,11 a new beam model was cre-
ated to improve dose predictions at extended SSD. For this
purpose, dose profiles for field sizes of 10 × 10 cm2 , 30
× 30 cm2 , and 40 × 40 cm2 at 100 cm SSD were utilized
from the original beam model as groundwork. The additional
inputs for the new model are inline, crossline, and depth dose
profiles that were taken with an IBA blue water phantom at
185 cm SSD. These dose profiles were for field sizes of 30 ×
30 cm2 and 40 × 40 cm2 , with crossline and inline scans at
the depths of 1.5, 10, and 20 cm. In addition to scans with the
tank centered on the field’s central axis, the water tank was
offset by ±15 cm in each direction to fully capture the tail
of the dose distribution for the beam field. Predictions for 13
and 7 ion chamber measurement points inside a child and a
baby anthropomorphic water phantom (see Sec. II.F), respec-
tively, were extracted with both beam models for comparison
to measurements. This allows the accuracy of the new beam
model to be compared to the previous one.
II.C. Measurement of floor backscatter
Due to the patient setup at extended SSD, scatter con-
ditions during TBI treatment are different from isocenter
treatments.12 The backscatter off the floor contributes to the
patient dose during treatment. For this reason, the relative
backscatter for a 6 MV photon beam was measured using
the following setup. With the gantry at 0◦ , an ion chamber
was mounted in the center of the beam field at 5, 10, 20, and
30 cm above the floor. Then, with the gantry rotated to 90◦ ,
the ion chamber was set up on the central axis of the beam
at the same distances from the source as in the previous mea-
surements and the same number of monitor units (MU) was
delivered using an open field size of 40 × 40 cm2 at isocenter.
The relative amount of backscatter, Brel , was calculated from
the measured dose D0 at 0◦ gantry angle and D90 at 90◦ gantry
angle, according to Eq. (1). This measurement was performed
with build-up materials to create an effective depth of 1.5 cm
for the ion chamber,
D0 − D90
Brel = . (1)
D90
II.D. Optimal treatment beam parameters
MATBI is possible for a wide variety of beam parame-
ters, such as beam geometry, field size, and total number of
beams. To find the optimal beam settings, several treatment
5265 Held et al.: Dosimetric aspects of modulated-arc TBI
plans have been calculated on patient CT scans. Each plan
was evaluated with respect to dose uniformity and treatment
time.
During patient simulation, radiopaque markers are placed
on the front and back of the patient, inline with the sternum.
The beams are aligned with the patient such that the field cen-
ter of the 0◦ gantry angle beam aligns with these setup points.
For treatment planning, a dosimetrist contours the outline of
the body contour and the lungs, which takes about 15 min on
average. The physician draws the contours for the lung blocks
and any additional organ at risk.
Prior to the optimization, beams were set up so that the
most outside ones contain at least 5 cm of the target in the
beam field at isocenter, which can help reduce the total MU
of the plan. Furthermore, the fields were distributed along the
arc with an angular offset between supine and prone beams
to avoid parallel-opposed beams. The dose for these beams is
calculated using the adaptive convolve algorithm in Pinnacle.
Thus, body inhomogeneity is corrected for. Inverse-planned
beam-weight optimization with a single constraint of uni-
form dose to the body contour was used to calculate an opti-
mal dose distribution. This optimization process takes roughly
15 min on up-to-date Pinnacle stations. To investigate the op-
timal field size, the field length covering the craniocaudal di-
rection of the patient was varied between 20, 30, and 40 cm.
The field width that covers the patient laterally was kept at
the maximum of 40 cm for all plans. Treatment times were
calculated and compared for each field size based on the total
amount of MU. Since there are data that suggest that lower
dose rates reduce pulmonary complications,13–17 the dose rate
for the beams that directly irradiate the lungs was set to
50 MU/min. To reduce treatment time, all other beams were
increased to a dose rate of 300 MU/min. In the next step, the
total number of beams was reduced from 45 to 23 by increas-
ing the gantry step size from 5◦ to 10◦ .
All treatment plans were evaluated based on the dose-
volume histogram (DVH) for the entire patient body. The pa-
tient used for this beam parameter test was prescribed with the
total dose of 200 cGy to the mean body dose. To quantify
the difference in the overall dose distribution, the fraction
of the volume within ±10% of the prescription dose, V(±10),
was extracted from each plan. The rationale for this number
was that the aim of TBI is to treat the whole body within
±10% of the prescribed dose.5 In addition, the standard de-
viation for the mean patient dose relative to their prescription
dose was extracted from the TPS.
II.E. Effect of beam setup errors
External markers are utilized to align the patient in the
beam field prior to every treatment fraction. Since the patient
is repositioned between supine and prone deliveries, misalign-
ment of these two beam sets is possible. To estimate the po-
tential errors from this, a misalignment in every direction has
been simulated on a patient CT scan in Pinnacle. Plans were
optimized as before. The lung blocks were included in dose
calculations, according to Sec. II.H. Then, the beam isocen-
ter for all prone beams was shifted with respect to that of the
Medical Physics, Vol. 39, No. 8, August 2012
5265
supine beams in all directions, i.e., 1, 5, and 10 cm in the
lateral, inferior, and anterior directions. The position of the
blocks relative to the patient remained the same. Each time,
the dose distribution was recomputed. This offset introduced
an error that corresponds to a patient misalignment between
supine and prone setup during treatment. Despite the 10 cm
shift, the body contour was still entirely covered by the treat-
ment beams. The impact on dose distribution within the body
is described by the V(±10) for the external body contour. To
indicate the impact of an isocenter shift on the dose distribu-
tion within the lungs, maximum, and mean doses are reported
for the left lung contour. The patient plan utilized for this test
had a prescription of 150 cGy per fraction to a percentage of
the mean body contour minus the skin, i.e., 5 mm.
II.F. Anthropomorphic water phantoms
Three water phantoms of different sizes were constructed
for the purpose of dose verification measurements and op-
timizing treatment planning parameters. The phantoms are
made from 0.64 cm thick acrylic sheets that were glued to-
gether to form boxes of different shapes to represent a patient
body. These three phantoms had the lengths of 71, 113, and
170 cm, which represent the height of a baby, a child, and an
adult, respectively. The adult phantom has ten footprints, la-
beled with numbers 1 through 10, for placement of an IBA
linear diode array [shown in Fig. 2(a)]. The array measured
the dose at a water depth of 5.0 cm in the arms, 8.1 cm in the
legs, and 10.1 cm in the torso. The other two phantoms use
acrylic blocks as mounts for an IBA cc13 ion chamber [see
Fig. 2(b)]. For the baby and child phantoms, these mounts
allow for 11 and 13 different point dose measurements, re-
spectively. These are labeled with the number 1 through 6,
to denote the ion chamber mount, and with an additional let-
ter s, i, l, or r to denote the superior, inferior, left, or right
(a)
(b)
F IG . 2. (a) The 170 cm tall anthropomorphic adult water phantom, showing
the 10 positions for the IBA linear diode array, which is placed in position 1.
(b) The 113 cm tall anthropomorphic child water phantom with five acrylic
blocks as ion chamber mounts that provide 13 different measurement posi-
tions. The torso and arms of this child phantom serve a dual purpose as the
torso and legs, respectively, of the baby phantom.
5266 Held et al.: Dosimetric aspects of modulated-arc TBI
orientation of the ion chamber in the mount. The ion cham-
ber was positioned at a water depth of 6.6 cm in the torso of
both phantoms. The dose was measured at 4.1 cm depth in the
legs of the baby model. There are no individual arms to this
phantom. Measurement points for the child phantom were at
4.4 cm water depth inside the legs and 4.1 cm inside the arms.
II.G. Dose delivery measurements
Absolute and relative dose measurements were performed
on the anthropomorphic water phantoms. In preparation, a CT
scan of each water phantom was acquired. The density at the
water surface was adjusted to eliminate the waves caused by
the CT table motion. An MATBI treatment plan was applied
to each phantom scan and the resulting dose distribution was
computed.
For the absolute dose point measurement, the child phan-
tom was placed on the treatment couch. All supine beams
of the MATBI treatment plan were delivered for each of the
13 measurement points. The point dose was measured with
an IBA cc13 ion chamber placed inside the acrylic mounts
and was averaged over both negative and positive electrom-
eter polarities. Absolute dose point measurements were also
performed for the baby phantom, which was built to fit in-
side a single open-field beam. To maximize the field length,
the collimators were rotated by 45◦ and one diamond-shaped
beam field with 40 × 40 cm2 field size at isocenter was used
to deliver the planned dose. This was repeated for seven cc13
positions throughout the phantom. The dose was again deliv-
ered for both polarities.
Relative dose measurements were performed with the adult
phantom. For each of the ten footprints, the supine beams for
an MATBI plan were delivered. The dose profile was mea-
sured with the IBA linear diode array that has 99 diodes,
spaced 0.5 cm apart.
II.H. Modeling of OAR blocks
To display the final dose distribution of each individ-
ual plan, cerrobend lung blocks, which are used during
the patient’s treatment, can be included in the optimized
treatment plan. For this reason, megavoltage cone-beam CT
(MVCBCT) images are acquired of the physical blocks af-
ter they are mounted onto the acrylic block holder bridge for
treatment. The use of MVCBCT enables the shape of the cer-
robend blocks to be acquired without significant imaging ar-
tifacts. This lung-block image is registered to the patient im-
age, at the intended location. Then, contours of the blocks
on the MVCBCT image are transferred to the patient image
and overridden to the cerrobend density of 9.3 g/cm3 . This
requires utilization of a density table in Pinnacle for high-
density materials. The dose calculation algorithms can lose
accuracy when high-density materials are present, so trans-
mission measurements were performed to ensure the validity
of these calculations. For that reason, the adult water phantom
was used for dose measurements. A previously imaged 1.7 cm
thick lung block, as it is used during treatment, was mounted
above the torso of the phantom, centered on the 0◦ gantry an-
Medical Physics, Vol. 39, No. 8, August 2012
5266
gle beam field. Then, the supine treatment for a patient was
delivered to the phantom and a dose profile was taken with
the linear diode array placed in footprint 4 [see Fig. 2(a)] at
13.8 cm below the block, which corresponded to a water
depth of 10.1 cm. Likewise, the blocks were included in the
TPS at the same position relative to the CT images. After
overriding the lung block density and computing the dose
distribution, the predicted dose profile for the diode array was
extracted from the treatment plan and compared to the mea-
sured values.
II.I. Effect of rotational delivery across blocks
The use of gantry rotation to deliver TBI blurs the dose dis-
tribution under the cerrobend blocks, causing an increase in
dose penumbra along the superior/inferior direction. The lung
blocks used during treatment are nondivergent, as each beam
irradiates the target from a different gantry angle. The blocks
are produced such that the partial lung volume is blocked with
a 1 cm margin inside the outline of the lungs. To quantify the
effect of the blurring, the dose distribution for rotational de-
livery across physical lung blocks has been studied in com-
parison to that for AP fields, using the ability of modeling
cerrobend blocks in Pinnacle. For this purpose, an MVCBCT
image of the blocks was fused with the CT scan images of the
adult phantom. The blocks were roughly centered on the 0◦
gantry angle beam field as they are during MATBI treatment.
As previously described, the block density was overridden
to 9.3 g/cm3 . Then, the dose distribution was computed for
an AP field and for an MATBI treatment, with gantry angles
between 295◦ and 35◦ . Additionally, the same computations
were performed in the absence of blocks. The transmission is
computed as the quotient of the dose with and without blocks.
This calculation was performed for the depths of 4, 8, 12, and
16 cm below the blocks.
III. RESULTS
III.A. Simulation
The dose distribution inside the Rando phantom was com-
pared by examining the calculated dose difference between
two CT scans with a CTDI of 20 mGy and 1.4 mGy. The first
scan had a 2 mm slice thickness, 300 mA tube current and
120 kVp tube voltage and the second had a 5 mm slice thick-
ness, 50 mA tube current and 90 kVp tube voltage. Due to the
differences in CT image resolution, large discrepancies ap-
pear at the transition between the phantom material and air. A
region of interest that was defined in the abdomen of the phan-
tom, away from these problematic areas, shows close agree-
ment between the scans. Within this region, the mean dose
difference was 0.14% and the largest one was 0.65%.
III.B. Measurements of dose delivery accuracy
For the child phantom, the relative differences between the
predictions and measurements for all cc13 positions were be-
tween −3.2 and 3.1% when using the new beam model and
5267 Held et al.: Dosimetric aspects of modulated-arc TBI 5267

Prediction Measurement

1.1 1 2

0.9

1.1 3 4

0.9
Relative Dose [%]

1.1 5 6

0.9

1.1 7 8

1
F IG . 4. Relative amount of backscattered radiation off the floor measured
0.9 with build-up material around the ion chamber.

1.1 9 10

These show that, when the gantry step size remains constant,
1
dose uniformity is best for the largest field size of 40 × 40
0.9 cm2 . For fields of 40 × 40 cm2 , 30 × 40 cm2 , and 20 × 40
−20 −10 0 10 20 −20 −10 0 10 20 cm2 size, V(±10) equals 89.5%, 87.9%, and 82.5%, respec-
Diode Position [cm]
tively. The corresponding standard deviation of the body dose
F IG . 3. Relative dose comparison for the predicted and measured dose, relative to the mean dose of 200 cGy is 8.7%, 9.0%, and 9.8%,
with the linear diode array for ten different positions inside the adult water respectively. Dose uniformity decreases with the number of
phantom. beams used during MATBI treatment. V(±10) is 89.5% for
45 treatment fields (5◦ step size) as opposed to 79.5% for 23
between −3.2 and 3.7% when using the old beam model for
beams (10◦ step size). The corresponding standard deviation
the dose computation. The new and old beam models pro-
relative to the mean dose is 8.2% and 9.5%, respectively. For
vided an average absolute difference of 1.2% and 1.4%, re-
a prescribed dose of 200 cGy, the estimated beam-on time for
spectively, from the measured values. Dose measurements for
MATBI supine and prone treatment is 67 min for a 190 cm tall
the seven positions inside the baby phantom were within −0.3
person with a 40 × 40 cm2 field size and a 50 MU/min dose
and 1.7% of the dose calculations for the new beam model
rate for all beams. Treatment time increases to 81 min and
and between −1.0 and 2.4% for the old beam model. For this
phantom, the new and old beam models had an average abso-
lute difference of 1.2% and 1.8%, respectively. Figure 3 shows
Volume Above Dose [%]

1
the profile for each diode array position inside the adult phan- 40 cm by 40 cm
30 cm by 40 cm
tom. The measured dose profiles inside the adult water phan- 20 cm by 40 cm
tom show undulating patterns along the array. The mean ab- 0.5
solute difference between the measured and calculated dose
for each of the ten footprints is less than 3%.
0
1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4
Dose [Gy]
III.C. Backscatter
Volume Above Dose [%]

1
Figure 4 shows the relative amount of backscatter mea- 45 beams
23 beams
sured at 5, 10, 20, and 30 cm above the floor with buildup ma-
terials wrapped around an ion chamber. The relative backscat- 0.5
ter decreases with increasing distance to the floor. At 20 cm
above the floor, which is roughly the distance of the treatment
couch top to the floor, the relative backscatter contribution re- 0
duces to 1.6%. 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4
Dose [Gy]

F IG . 5. Dose-volume histograms demonstrating the effect of beam param-

III.D. Treatment beam parameters eters on target coverage. (a) The dose uniformity for different field sizes
at constant gantry step size of 5◦ . (b) Dose uniformity for different num-
The DVHs of Fig. 5 display the effect of the different field ber of beams used during treatment, where all beams have a field size of
sizes (a) and the number of beams (b) on the dose uniformity. 40 × 40 cm2 .

Medical Physics, Vol. 39, No. 8, August 2012

5268 Held et al.: Dosimetric aspects of modulated-arc TBI 5268

1 TABLE II. Change of the maximum and mean dose inside the left lung con-
no shift tour for a 1, 5, and 10 cm shift into lateral, craniocaudal, and SSD direction
0.9 10 cm lateral relative to the original plan in percent. For no shift, the maximum dose is
10 cm superior
167 cGy and the mean dose is 116 cGy per fraction inside the left lung
0.8 10 cm shorter SSD
contour.
Volume Above Dose [%]

0.7
Relative dose difference in lungs (%)
0.6
Max Mean
0.5
Setup error (cm) Lateral Craniocaudal SSD Lateral Craniocaudal SSD
0.4
1 −0.13 −0.19 −0.68 −0.02 −0.10 −0.69
0.3 5 −1.50 −0.13 −5.77 −0.33 −0.82 −3.65
0.2 10 −2.34 −0.54 −14.39 −0.70 −0.85 −7.81

0.1

0 of 1, 5, and 10 cm. The standard deviation relative to the body

1.2 1.3 1.4 1.5 1.6
Dose [Gy]
F IG . 6. Dose-volume histogram for an optimized MATBI plan, compared to
plans that contain an offset of the anterior isocenter by 10 cm to all directions.
95 min for the field sizes of 30 × 40 cm2 and 20 × 40 cm2 ,
respectively. This shows that the best treatment times are also
achieved with the maximum field size of 40 × 40 cm2 at
isocenter. The beam-on time is reduced even further when us-
ing a dose rate of 300 MU/min for beams that do not interfere
directly with the lungs. For this, treatment times are estimated
to be 29 min for a 40 × 40 cm2 field size and 32 min for both
30 × 40 cm2 and 20 × 40 cm2 . The smaller field sizes have
fewer beams interfering with the lung; thus, more beams can
be delivered with the high dose rate to 300 MU/min. Conse-
quently, the treatment time reduction for utilizing the largest
field size becomes less significant.
III.E. Effect of beam setup errors
Figure 6 shows the overall coverage of the target volume
for each plan with an introduced offset in isocenter position
compared to the dose distribution obtained from the original
plan. This figure shows that a 10 cm shift to inferior, supe-
rior, or lateral direction has a minimal impact on the overall
dose coverage of the body. On the contrary, a severe error
in dose delivery is introduced when misaligning the patient
vertically. The change in SSD causes the overall dose to in-
crease for smaller SSD and to decrease for larger SSD. For
the plan without beam offset, V(±10) is 79% and the stan-
1.7 1.8 1.9 2
mean dose of each is unchanged 11.9% for lateral and cranio-
caudal shifts. For a reduced SSD shift, the standard deviation
relative to the mean dose increases to 12.5%. This shift causes
severe increase in dose that reduces V(±10) to 26% for 10 cm
shorter SSD. Table II shows the maximum and mean dose in-
side the left lung contour in percent for 1, 5, and 10 cm setup
error relative to the initial plan. Lateral and craniocaudal shifts
show no difference in the mean dose (<1%) but increase the
maximum dose (<3%). The largest impact on the dose dis-
tribution is introduced by a reduced SSD. In that case, the
relative maximum and mean dose increases as much as 14%
and 8%, respectively, for a 10 cm shift.
III.F. Modeling of OAR blocks
Figure 7 displays the measured and calculated dose pro-
files underneath the OAR block. Both lines correlate well.
Between the distances of 55 cm and 95 cm along the diode
array, the block attenuates the dose, showing less transmitted
dose in this area. The average dose below the block is calcu-
lated for the distance between 68 cm and 83 cm on the array
relative to the dose outside the range of the block, between 0
and 20 cm. The measurement shows an average relative dose
Prediction
Measurement
1.1
1
Relative Dose

dard deviation relative to the mean dose is 11.9%. Table I lists 0.9
the V(±10) for lateral, craniocaudal and decreased SSD shifts
0.8

TABLE I. Amount of body volume that is within ±10% of the prescribed 0.7
dose when introducing a setup error of 1, 5, and 10 cm to the prone isocenter.
0.6
Body volume inside ±10% of prescribed dose (%)

Setup error (cm) Lateral Craniocaudal SSD 0.5

0 20 40 60 80 100
1 78.55 78.54 75.72 Distance [cm]
5 78.92 78.55 57.15
10 79.88 78.78 25.66 F IG . 7. Predicted and measured dose profile underneath a physical lung
block.

Medical Physics, Vol. 39, No. 8, August 2012

5269 Held et al.: Dosimetric aspects of modulated-arc TBI 5269

4 cm 8 cm
1.1 1.1

1 1

Transmission [%]

Transmission [%]
0.9 0.9

0.8 0.8

0.7 0.7

0.6 0.6

0.5 0.5

0.4 0.4
10 5 0 5 10 10 5 0 5 10
Length [cm] Length [cm]

12 cm 16 cm
1.1 1.1

1 1
Transmission [%]

Transmission [%]
0.9 0.9

0.8 0.8

0.7 0.7

0.6 0.6

0.5 0.5

0.4 0.4
10 5 0 5 10 10 5 0 5 10
Length [cm] Length [cm]
straight on field arc fields

F IG . 8. Dose profile predictions for transmitted dose at different distances underneath a physical lung block.

of 55% compared to a predicted average relative dose of 57%
below the block.
III.G. Effect of rotational delivery across blocks
Figure 8 displays the calculated dose transmission profile,
which is the quotient of the dose inside the phantom with and
without blocks, along the superior-inferior direction at differ-
ent depths. Profiles are calculated for distances of 4, 8, 12, and
16 cm below the block. The angular beam incidence causes
dose to reach into the region underneath the block, thus in-
creasing the penumbra with depth. The penumbra is defined
here as the distance between the 90 and 60% transmission
values. It is measured for the inferior and superior edge of the
block and is then averaged. At 4 cm, the field penumbra is
1.6 cm for rotational dose delivery as opposed to 0.6 cm for
AP fields. At 16 cm depth, the corresponding penumbra in-
creases to 5 cm and 0.8 cm, respectively.
IV. DISCUSSION
Although lower scan values reduce the CT image qual-
ity, this has no significant effect on the accuracy of treatment
planning and dose computation. Thus, the amount of data and
the time required for contouring can be reduced for scans us-
ing larger slice thickness and lower scan values. Acquiring
fewer images per patient also minimizes dose to the patient.
However, compared to the prescribed dose, which the patient
will receive during treatment, the reduction in dose to the pa-
tient during the simulation is clinically irrelevant.
A new beam model was created specifically for the
extended-SSD patient position of MATBI to investigate the
potential dose accuracy improvements. One challenge in cre-
ating a suitable beam model for MATBI is obtaining suffi-
cient accuracy in the tail of the dose distribution. Since the
radiation field is swept along the full length of the patient,
each point of the anatomy receives dose from the tail of many
different beams. Thus, small errors in the dose-tail modeling
can accumulate to make a large discrepancy. The new beam
model focused on this issue. Nevertheless, in this case, the
new MATBI beam model shows no real significant improve-
ment on the average absolute dose error. However, the benefits
of developing an MATBI-specific beam model would vary at
different institutions,11, 18 as the accuracy of a beam model in
the dose tail depends not only on the person commissioning
the model but also the TPS itself.
Another consequence of MATBI treatment compared to
isocentric irradiation is the presence of additional backscat-
tered dose off the floor. Measurements show that the couch
height of 20 cm above the floor reduces the relative amount
of backscatter to 1.6%. This together with the good agree-
ment between dose calculations and ion chamber measure-
ments was sufficient evidence to neglect backscatter off the
floor.

Medical Physics, Vol. 39, No. 8, August 2012

5270 Held et al.: Dosimetric aspects of modulated-arc TBI
The field size of 40 × 40 cm2 was optimal not only for
treatment time, but also dose homogeneity. This might appear
counter intuitive, as one would expect smaller field sizes to
allow for better dose modulation. However, since the gantry
angle step size was fixed, using large beam fields causes more
beams to overlap at each point. The end of one beam along the
patient length does not exactly coincide with the beginning of
another. This introduces a dose undulation with a relative am-
plitude that is inversely proportional to the number of overlap-
ping beams. Thus, the higher number of overlapping beams
with a 40 × 40 cm2 field size reduces this amplitude, and
therefore also improves homogeneity. Regions of high dose
(>10% above the prescription dose) appear within the arms
and fingers. Since MATBI does not modulate the dose later-
ally, these anatomical structures, which are thin compared to
the torso, receive a higher dose than the prescription dose. The
highest dose points appear at the peak of the dose undulations
in the arms and fingers.
In addition to these dose undulations, the intermittent dose
delivery for TBI causes the potential problem that malignant
cells might evade irradiation due to cell circulation. Molloy19
has investigated this issue in detail, using the field and treat-
ment time parameters of tomotherapy-based TBI treatment.
With reference to these results, the effect of intermittent dose
delivery is even less of an issue for MATBI than it was for the
results in Molloy. MATBI treatment times are long enough to
reduce the effect of the intermittency. Furthermore, each field
irradiates about 80 cm along the axial length of the patient
and therefore covers a sufficient part of the body at any given
time.
The total-body dose coverage is not easily susceptible to
errors from misalignments of the patient in the lateral or
superior-inferior directions. This alignment is, however, more
critical for organs at risk. Maintaining the correct SSD is cru-
cial to the dose for both the total body and the organs at risk.
The V(±10) itself is not sufficient to reflect errors in the dose
distribution caused by misalignment. On the contrary, a slight
increase of the V(±10) values for lateral shifts (see Table I)
could be misleading. In addition, the standard deviation and
an evaluation of the overall DVH are important for a detailed
description of dose errors.
The comparison of measurements and dose calculations
shows that Pinnacle can model the presence of high-density
materials for lung blocks. Thus, the effect of these blocks
can be included into patient dose calculations. The ability to
model the OAR blocks was utilized to investigate the effect
of rotational delivery across physical lung blocks. This in-
vestigation demonstrates that compared to AP beams MATBI
yields similar minimum block transmission values, but that
the dose penumbra increases with depth. For this reason, it is
important to place the blocks as close to the patient as possi-
ble, ideally onto the patient’s skin.
V. CONCLUSION
The modulated-arc total-body irradiation technique em-
phasizes the delivery of a uniform dose distribution to the
Medical Physics, Vol. 39, No. 8, August 2012
5270
patient in a comfortable and stable position in standard-sized
vaults. The CT-based inverse planning approach allows dose
distribution optimization and evaluation. The MATBI beam
parameters have a large influence on the treatment time and
also affect dose uniformity. Consequently, the application of
optimum beam settings that are presented here is crucial when
implementing MATBI. This study also lays out the required
measurements to ensure the accuracy of MATBI dose de-
livery. Both of these dosimetric aspects are essential for the
clinical implementation of MATBI. The technique has al-
ready been used on 23 patients and is now the standard at
UCSF.
ACKNOWLEDGMENTS
This work was supported in part by Siemens.
a) Author to whom correspondence should be addressed. Electronic mail:
jpouliot@radonc.ucsf.edu; Telephone: (415) 353-7190; Fax: (415) 353-
9883.
1 J. C. Breneman, H. R. Elson, R. Little, M. Lamba, A. E. Foster, and
B. S. Aron, “A technique for delivery of total body irradiation for bone
marrow transplantation in adults and adolescents,” Int. J. Radiat. Oncol.,
Biol., Phys. 18, 1233–1236 (1997).
2 S. V. Harden, D. S. Routsis, A. R. Geater, S. J. Thomas, P. J. Taylor,
R. E. Marcus, and M. V. Williams, “Total body irradiation using a modified
standing technique: A single institution 7 year experience,” Br. J. Radiol.
74, 1041–1047 (2011).
3 C.-S. Chui, D. P. Fontenla, E. Mullokandov, A. Kapulsky, Y.-C. Lo,
and C.-L. Lo, “Total body irradiation with an arc and a gravity-
oriented compensator,” Int. J. Radiat. Oncol., Biol., Phys. 39, 1191–1195
(1997).
4 M. Chrétien, C. Côté, R. Blais, L. Brouard, L. Roy-Lacroix, and
M. Larochelle, “A variable speed translating couch technique for total body
irradiation,” Med. Phys. 27, 1127–1130 (2000).
5 M. Sarfaraz, C. Yu, D. J. Chen, and L. Der, “A translational couch tech-
nique for total body irradiation,” J. Appl. Clin. Med. Phys. 2, 201–209
(2001).
6 A. Hussain, J. E. Villarreal-Barajas, P. Dunscombe, and D. W. Brown,
“Aperture modulated, translating bed total body irradiation,” Med. Phys.
38(2), 932–941 (2011).
7 M. Held, “A new modulated-arc inverse-planned total body irradiation
technique,” Diploma thesis, Department of Physics, University of Ham-
burg, Hamburg, Germany, 2011.
8 N. Kirby, “Inverse-planned modulated-arc total-body irradiation,” Med.
Phys. 39, 2761–2764 (2012).
9 T. B. Shope, R. M. Gagne, and G. C. Johnson, “A method for describing the
doses delivered by transmission x-ray computed tomography,” Med. Phys.
8, 488–495 (1981).
10 H. D. Nagel, Radiation Exposure in Computed Tomography (COCIR,
Frankfurt, 2000).
11 M.-C. Lavallée, L. Gingras, M. Chrétien, S. Aubin, C. Côté, and
L. Beaulieu, “Commissioning and evaluation of an extended SSD photon
model for PINNACLE3 : An application to total body irradiation,” Med.
Phys. 36, 3844–3855 (2009).
12 J. Van Dam, A. Rijnders, L. Vanuytsel, and H.-Z. Zhang, “Practical
implication of backscatter from outside the patient on the dose distri-
bution during total body irradiation,” Radiother. Oncol. 13, 193–201
(1988).
13 J. Van Dyk, T. J. Keane, S. Kan, W. D. Rider, and C. H. J. Fryer, “Radiation
pneumonitis following large single dose irradiation: A single re-evaluation
based on absolute dose to lung,” Int. J. Radiat. Oncol., Biol., Phys 7(4),
461–467 (1981).
14 M. M. Bortin, H. E. M. Kay, R. P. Gale, and A. A. Rimm, “Factors asso-
ciated with interstitial pneumonitis after bone-marrow transplantation for
acute leukaemia,” Lancet 319(8269), 437–439 (1982).
5271 Held et al.: Dosimetric aspects of modulated-arc TBI
15 R. S. Weiner, M. M. Bortin, R. P. Gale, E. Gluckman, H. E. Kay, H. J. Kolb,
A. J. Hartz, and A. A. Rimm, “Interstitial pneumonitis after bone marrow
transplantation,” Ann. Intern. Med. 104(2), 168–175 (1986).
16 C. A. Barker, T. J. LoSasso, and S. L. Wolden, “Total body irradiation,”
Textbook of Radiation Oncology, edited by T. L. Phillips, R. T. Hoppe, and
M. Roach III (W. B. Saunders, Philadelphia, 2004).
17 M. Beyzadeoglu, K. Oysul, B Dirican, F. Arpaci, A Balkan, S. Surenkok,
and Y. Pak, “Effect of dose-rate and lung dose in total body irradiation on
Medical Physics, Vol. 39, No. 8, August 2012
5271
interstitial pneumonitis after bone marrow transplantation,” Tohoku J. Exp.
Med. 202, 255–263 (2004).
18 A. Hussain, E. Villarreal-Baraja, D. Brown, and P. Dunscombe, “Validation
of the eclipse AAA algorithm at extended SSD,” J. Appl. Clin. Med. Phys.
11, 90–100 (2010).
19 J. A. Molloy, “Statistical analysis of dose heterogeneity in circulating
blood: Implications for sequential methods of total body irradiation,” Med.
Phys. 37, 5568–5578 (2010).