Physics in Medicine & Biology

PAPER Related content

- Modulation index for VMAT considering
Error detection capability of a novel transmission both mechanical and dose calculation
uncertainties
detector: a validation study for online VMAT Jong Min Park, So-Yeon Park and
Hyoungnyoun Kim

monitoring - Clinical evaluation of a novel transmission

detector for 3D quality assurance of IMRT
and SBRT
To cite this article: Marlies Pasler et al 2017 Phys. Med. Biol. 62 7440 Gueorgui Gueorguiev, Fazal Khan, Dolla
Toomeh et al.

- Verification of VMAT treatment plans

Ramesh Boggula, Mattias Birkner, Frank
Lohr et al.
View the article online for updates and enhancements.

This content was downloaded from IP address 141.48.25.3 on 11/01/2018 at 15:24

 Institute of Physics and Engineering in Medicine Physics in Medicine & Biology

Phys. Med. Biol. 62 (2017) 7440–7450 https://doi.org/10.1088/1361-6560/aa7dc7

Error detection capability of a novel

transmission detector: a validation study
for online VMAT monitoring
Marlies Pasler1,5, Kilian Michel1,2, Livia Marrazzo3,
Michael Obenland1, Stefania Pallotta3,4, Mari Björnsgard1
and Johannes Lutterbach1
1
  Lake Constance Radiation Oncology Center, Singen, Friedrichshafen, Germany
2
  Naturwissenschaftliche Fakultät II, Martin-Luther-Universität, Halle-Wittenberg,
Germany
3
  Medical Physics Unit AOU Careggi, Florence, Italy
4
  Department of Biomedical, Experimental and Clinical Sciences—University
of Florence, Florence, Italy

E-mail: pasler@strahlentherapie-fn.de

Received 4 April 2017, revised 3 July 2017

Accepted for publication 5 July 2017
Published 1 September 2017

Abstract
The purpose of this study was to characterize a new single large-area ionization
chamber, the integral quality monitor system (iRT, Germany), for online and
real-time beam monitoring.
Signal stability, monitor unit (MU) linearity and dose rate dependence
were investigated for static and arc deliveries and compared to independent
ionization chamber measurements. The dose verification capability of the
transmission detector system was evaluated by comparing calculated and
measured detector signals for 15 volumetric modulated arc therapy plans.
The error detection sensitivity was tested by introducing MLC position and
linac output errors. Deviations in dose distributions between the original and
error-induced plans were compared in terms of detector signal deviation,
dose-volume histogram (DVH) metrics and 2D γ-evaluation (2%/2 mm and
3%/3 mm).
The detector signal is linearly dependent on linac output and shows
negligible (<0.4%) dose rate dependence up to 460 MU min−1. Signal
stability is within 1% for cumulative detector output; substantial variations
were observed for the segment-by-segment signal.

5
Author to whom any correspondence should be addressed.

1361-6560/17/187440+11$33.00  © 2017 Institute of Physics and Engineering in Medicine  Printed in the UK 7440
Phys. Med. Biol. 62 (2017) 7440 M Pasler et al

Calculated versus measured cumulative signal deviations ranged

from  −0.16%–2.25%. DVH, mean 2D γ-value and detector signal evaluations
showed increasing deviations with regard to the respective reference with
growing MLC and dose output errors; good correlation between DVH
metrics and detector signal deviation was found (e.g. PTV Dmean: R2  =  0.97).
Positional MLC errors of 1 mm and errors in linac output of 2% were identified
with the transmission detector system.
The extensive tests performed in this investigation show that the new
transmission detector provides a stable and sensitive cumulative signal output
and is suitable for beam monitoring during patient treatment.

Keywords: intra-fraction linac monitoring, online dose verification, transmission

detector, quality assurance, volumetric modulated arc therapy (VMAT)

S Supplementary material for this article is available online

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

1. Introduction

Volumetric modulated arc therapy (VMAT) requires a thorough and independent quality
assurance (QA) program to ensure reliable, stable and reproducible plan delivery (Yu et al
2011). Rotational techniques are interesting in online beam monitoring because it is increas-
ingly important to verify the precise interplay and coordination of dynamic linac components
during plan delivery. While pre-treatment QA verifies dose delivery feasibility and accuracy
of a system, potential treatment delivery errors occurring in subsequent fractions could still
remain undetected. Moreover, online adaptive radiotherapy approaches require the integration
of online dose verification to ensure patient safety. Therefore, the importance of online and
real-time patient dose verification and beam monitoring systems has increased over the years
(Fuangrod et al 2016, Spreeuw et al 2016, Thoelking et al 2016).
EPID dosimetry and 2D detector arrays play a major role in patient specific QA (Mans et al
2010, Riley et al 2011, Olaciregui-Ruiz et al 2013). Recently, EPID based dose measurements
proved to be fast enough to be employed in real-time for VMAT (e.g. Spreeuw et al 2016).
Transmission detectors, placed in the beam path between the treatment head and the patient,
enable accurate verification of dose delivery and linac monitoring. While measurements are
performed online, the actual verification is typically done offline (Wong et al 2012, Thoelking
et al 2016). Hence, serious overdosage, which cannot be corrected for in subsequent fractions
by replanning, could still occur.
Recently, a new transmission detector consisting of a single large-area ionization chamber
became commercially available that allows real-time online beam monitoring (Islam et al 2009).
The system displays the measured signal during patient treatment and thus facilitates real online
verification rather than offline verification. In recent studies, the transmission detector was charac-
terized with regard to its influence on photon beam fluence, response to dose and the accuracy of
the integrated barometer, thermometer and inclinometer (Casar et al 2017, Hoffmann et al 2017).
Neither of the above mentioned studies addressed the dose verification capability for
VMAT, nor do they provide extensive investigation of error detection sensitivity and control
point resolved beam monitoring for VMAT deliveries.
The aim of the present work was to investigate the clinical performance of the new system
by means of dosimetric characteristics during rotation, accuracy of signal prediction, error

7441
Phys. Med. Biol. 62 (2017) 7440 M Pasler et al

Figure 1. IQM attached to the linac head (a) and schematic illustration (b) of the
IQM detector (courtesy of iRT systems). Spatial sensitivity of the measured signal is
provided by the alignment of three electrodes: two outer electrodes are angled relative
to the central collecting electrode.

detection capability and segment-by-segment beam monitoring. Furthermore, data is provided

to be used as a baseline for benchmarking patient specific linac QA and an error tolerance
level is proposed.

2.  Methods and materials

2.1.  Transmission detector

The integral quality monitor system (IQM, iRT, Germany) consists of a single large-area
(26  ×  26 cm2) ionization chamber positioned between the linac head and the patient, as previously
described in detail by Islam et al (2009). The ionization chamber has a gradient along the direction
of the MLC motion and therefore features a spatially dependent signal sensitivity (figure 1).
Gantry and collimator angles are monitored using an integrated inclinometer. The data
acquisition software of the detector system is interfaced with the linac to access patient spe-
cific treatment parameter information; including patient ID, field name and field monitor units
(MU). The communication between the IQM system and the linac (Elekta synergy) is estab-
lished via Elekta’s iCom interface.
The detector provides a signal count for each single control point (segment-by-segment) and
a cumulative output in near real-time, allowing the detection of delivery deviations during the
treatment. For the segment-by-segment signal, the detector counts are cumulated between two
consecutive control points, where the gantry angle is used to determine control point boundaries.
The dose verification is based on the comparison of a measured and a calculated signal
which is proportional to a spatially dependent dose-area-product of the beam. The calcu-
lated signal is derived from field parameters including the jaw settings, beam segment shapes
and monitor units. The calculation method takes into account various predetermined factors,
including the spatial response function of the detector, MLC characteristics, secondary jaw
transmission and field size factors; a detailed description is given in the work Islam et  al
(2009). In this work, a pre-clinical release of the calculation module is used (version 1.3.0-
beta) and evaluated. The reference signal is user-selectable: for dose verification, the calcu-
lated signal is defined as reference and a reference measurement is used to check consistency.

2.2.  Dosimetric quantities and signal stability

Dose rate dependence was measured for a range of clinical dose rates from 15 to 460
MU min−1. The system was not tested on dose rates higher than 460 MU min−1 due to unavail-
ability on our linac. MU linearity was tested for 3–1000 MU. Measurements were performed

7442
Phys. Med. Biol. 62 (2017) 7440 M Pasler et al

for both static beams and arcs and compared with ionization chamber measurements (Type
31013 PTW semiflex, volume 0.3 cm3) placed centrally in a 30  ×  30  ×  20 cm3 solid water
slab phantom.
Response reproducibility was tested over a period of three months for a static 10  ×  10 cm2
field, a step-and-shoot test-IMRT QA plan and a VMAT plan (HN-1, supplementary table 1
(stacks.iop.org/PMB/62/7440/mmedia)). The IMRT QA plan was part of the acceptance test-
ing and covers a large part of the detector area with spatially variable 4  ×  4 cm2-segments
(supplementary figure 1).
All measurements were repeated three times to evaluate standard deviation.

2.3.  Treatment plans and signal calculation

The new detector system was investigated by irradiating 15 randomly chosen VMAT plans
(supplementary table 1): five dual-arc head and neck (HN), five single-arc prostate (PC) plans
and five partial-arc breast cancer plans with a simultaneously integrated boost (MC). All plans
were generated with Pinnacle (V.14) for an Elekta synergy linac (MLCi2) with 6 MV photon
beam energy and 4° gantry spacing.
For the detector signal calculation, RTplan-files were transferred to the IQM calculator.
The calculated reference was compared to the measured signal. The modulation complexity
score (MCS), introduced by McNiven et al (2010), that incorporates variability in leaf posi-
tions, degree of irregularity in field shape, segment weight and aperture area was calculated
for each plan to investigate potential correlations with signal calculation accuracy. The score
decreases from a value of 1.0 if the complexity of the plan increases.

2.4.  Error detection capability

Basic data transfer errors and beam interruptions were simulated to check the IQM monitor-
ing system’s ability to correctly display potential malfunctions: a treatment plan with correct
patient and plan ID but (i) wrong number of MU, (ii) 10 MV instead of 6 MV photon beam
energy (iii) incorrect gantry angle or (iv) wrong collimator angle was tested.
Different types of errors for MLC positions and linac output were introduced to investigate
the error detection sensitivity of the transmission detector system. Measurement signals of
unmodified treatment plans were defined as a reference and compared to those of the error
plans. Geometric errors included shifts of both leaf banks (×1  ×  2) for all control points
toward (i) and away (ii) from the central axis of the beam and unidirectional shifts of both
leaf banks (iii) by 1 and 2 mm, respectively. Dosimetric errors were induced by increasing the
number of MUs by 2% and 5%.
Deviations in dose distributions between the original and error-induced plans were com-
pared in terms of IQM signal deviation for both segment-by-segment output and cumulative
output.

2.5. Evaluation

In order to estimate the clinical impact of the introduced errors, DVHs of unmodified plans
were compared with the modified plans using TPS information. DVH metrics were assessed
and evaluated manually: Dmean, D2% and D98% for planning target volumes (PTV) and organs
at risk (OAR).

7443
Phys. Med. Biol. 62 (2017) 7440 M Pasler et al

Figure 2.  IQM signal linearity for a 10  ×  10 cm2 field (a), dose rate dependence (b)
and cumulative signal reproducibility (c). Data present average values and standard
deviations (error bars). Abbreviations: IC  =  ionization chamber (Type 31013 PTW
semiflex, volume 0.3 cm3); IQM  =  integral quality monitor; QA IMRT  =  step-and-
shoot IMRT plan for quality assurance (QA), see supplementary figure 1; VMAT HN-1
a/b  =  volumetric modulated arc plan for head and neck (2 arcs).

The IQM signal deviations for the error induced plans were compared to a conventional
2D pre-treatment verification method. For the latter, the Octavius and the 2D ion chamber
array (seven29, PTW-Freiburg, Germany) (Spezi et al 2005) was used, as it is the equipment
for patient specific QA in our department. Planar dose distributions were calculated in the
horizontal plane for original- and error-plans. 2D gamma (γ) analysis was performed for the
γ-criteria γ2%/2 mm and γ3%/3 mm (dose difference/distance to agreement) and global gamma
evaluation with a dose threshold of 5%. Results are presented as mean  ±1 standard deviation
(SD).

3. Results

3.1.  Dosimetric quantities

The IQM signal is linearly dependent on MU delivered for both static (R2  =  1) and rotational
(R2  =  1) 10  ×  10 cm2 fields as shown in figure 2(a).
The dose rate dependence measurements are shown in figure 2(b). Measurements are nor­
malized to the value obtained at the maximum dose rate (460 MU min−1). The maximum IQM
signal difference observed for different dose rates is 0.16% and 0.38% for static delivery and
arc delivery, respectively. Similar results were observed for the ion chamber measurements
(figure 2).

3.2.  Signal stability

Figure 2(c) shows detector signal deviations normalized to the first measurement as a function
of time. The cumulative signal reproducibility is within 1% for all test beams. Based on the
measured signal stability for the VMAT plan (mean deviation from reference 0.47  ±  0.17%),
the tolerance level was defined as mean value  +3SD, i.e. 1.0%.
The segment-by-segment signal proved to be very stable for the static IMRT QA plan
(deviation  <  1%). However, in VMAT plans, segment-by-segment response varied substanti­
ally: up to threefold signal increase compared to the reference measurement was recorded,
indicating poor reproducibility. The segment-by-segment signal was not further investigated
for error detection capability because of its intrinsic limitations.

7444
Phys. Med. Biol. 62 (2017) 7440 M Pasler et al

3.3.  Calculated versus measured transmission detector signal

The agreement between signal calculation and measurement was very good for the cumulative
signal, the average deviation was 0.96  ±  0.34% (table 1). 2D gamma array measurements did
not correlate with IQM signal deviation (R2  =  0.04).
However, segment-by-segment signal analysis revealed large deviations for all three groups
of treatment plans: 4.4  ±  26.6%, 4.9  ±  19.4% and 8.2  ±  29.7% mean difference (±SD) were
found between calculation and measurement for PC, MC and HN plans, respectively, indicat-
ing high intersegment variability.
No correlation was found for MCS and agreement between signal calculation and measure-
ment (R2  =  0.24).

3.4.  Error detection capability

The IQM system is able to detect a delivery interruption and to resume the measurement
without user interaction. The deviation between the interrupted and uninterrupted cumulative
detector signal was  <0.5%. Transferring a treatment plan with the wrong number of MUs,
incorrect gantry or collimator angle to the linac causes an immediate alert before the actual
start of plan delivery due to disagreement of plan information in the R&V system and the data
stored in the IQM monitoring software. Delivering a VMAT plan with 10 MV instead of 6 MV
photon beam energy caused a signal decrease of 6.3  ±  1.5%.
Concerning the simulation of geometrical errors and linac output errors, DVH, mean 2D
gamma and IQM signal evaluations showed increasing deviations with regard to the respective
reference with growing MLC shifts and dose output deviations (figures 3 and 4). All recorded
DVH parameter deviations between the original and error plans for PTVs and OARs are listed
in supplementary table 2. DVH and IQM signal deviations of error plans with respect to the
original plan are illustrated in figure 3.
The cumulative detector signal deviated by up to 17.9% when comparing the original plan
to the error plan with 2 mm MLC shift with respect to the central beam axis. On average, a
1 mm shift of MLCs towards and away from the central axis resulted in a signal deviation
of  −5.24  ±  1.2% and 5.5  ±  1.2%, respectively. A 2 mm MLC shift doubled the IQM signal
deviation, as figures 3 and 4(b) illustrate.
Accordingly, for a 2 mm leaf bank shift towards or away from the central axis, the largest
γ-values were observed for the 2D array dose evaluation: the mean γ-value increased by up
to 1.54 for the 2%/2 mm criterion and up to 0.98 for the 3%/3 mm criterion, respectively. The
γ passing rate (%γ  <  1) decreased to 64.7% and 47.2% for the 3%/3 mm and 2%/2 mm crite-
rion, respectively. Figure 4 illustrates the correlation of IQM signal deviation with and DVH
metrics ((a), (c) and (d)) and γ-evaluation ((e) and (f)), respectively.
Unidirectional leaf bank shifts had very little effect on DVH parameters, e.g. the difference
in Dmean (PTV) was  <1% for a 2 mm shift of both leaf banks. Accordingly, the effect of unilat-
eral MLC shifts on dosimetric verification with the IQM system was negligible:  <1% signal
difference for the IQM was found, i.e. below the above defined tolerance.
For γ-evaluation, γ-values below 0.3 and 0.4 as well as %γ  <  1 greater than 99% and 95%
for 3%/3 mm and 2%/2 mm were determined, respectively. Detailed data on unidirectional
shifts is provided in supplementary figure 2.
With respect to linac output, the transmission detector signal correlated linearly with the
increase of 2% and 5%, as shown in figure 3. Accordingly, the 2% linac output increase resulted
in a γmean of 0.24  ±  0.08 and 0.36  ±  0.08 and %γ  <  1 of 98.7% and 96.1% for 3%/3 mm and
2%/2 mm, respectively; a 5% linac output increase resulted in a γmean of 0.52  ±  0.14 and
0.78  ±  0.22 and %γ  <  1 of 80.2% and 72.6% for 3%/3 mm and 2%/2 mm, respectively.
7445
Phys. Med. Biol. 62 (2017) 7440 M Pasler et al

Table 1. Cumulative measured IQM signal (signal counts, arb. unit), difference in
(%) between calculated and measured cumulative signal for all investigated rotational
fields, minimum (min) and maximum (max) of deviation between calculation and
measurement for the segment-by-segment signal, mean y-values and passing rates from
2D array measurements.
IQM 2D array
Cumulative
signal Range of
deviation segment-by-
from calc. segment signal
Measured ref. (%) deviation (%) 3%/3 mm 2%/2 mm
cumulative
signal Passing rate Passing rate
Patient (arb. unit) Mean  ±  SD Min Max (%γ  <  1) γmean (%γ  <  1) γmean
PC-1 315 558 0.2  ±  0.4 −36.8 −132.9 100.0 0.30 97.3 0.45
PC-2 423 100 0.7  ±  0.4 −47.3 −243.5 96.0 0.26 95.0 0.40
PC-3 416 912 0.2  ±  0.4 −45.2 −206.0 100.0 0.17 98.6 0.25
PC-4 490 688 0.1  ±  0.5 −38.3 −222.6 100.0 0.29 94.1 0.44
PC-5 507 645 0.5  ±  0.3 −42.6 −132.1 100.0 0.30 92.4 0.45
MC-1 918 985 0.2  ±  0.3 −34.4 −136.7 97.2 0.35 88.2 0.51
MC-2 850 226 2.3  ±  0.2 −30.8 −208.5 98.5 0.30 90.8 0.44
MC-3 913 392 1.0  ±  0.2 81.5 −141.7 99.4 0.25 97.2 0.37
MC-4 685 218 −0.2  ±  0.2 −91.5 −86.4 100.0 0.27 98.1 0.40
MC-5 974 701 1.9  ±  0.3 −28.9 −111.7 98.9 0.30 92.4 0.44
HN-1 a 674 684 0.6  ±  0.1 −40.6 −209.2 99.5 0.33 93.7 0.49
HN-1 b 632 022 1.1  ±  0.1 −38.4 −165.0 100.0 0.25 96.8 0.37
HN-2 a 415 114 2.0  ±  0.3 −38.6 −236.6 98.8 0.28 93.4 0.41
HN-2 b 653 930 1.3  ±  0.4 −59.6 −173.7 99.5 0.26 95.9 0.38
HN-3 a 483 247 1.8  ±  0.4 −58.7 −327.6 99.3 0.30 92.4 0.44
HN-3 b 363 975 1.1  ±  0.4 −49.8 −218.0 98.3 0.29 91.7 0.42
HN-4 a 502 184 2.0  ±  0.4 −38.3 −114.4 99.2 0.28 94.1 0.41
HN-4 b 333 586 1.0  ±  0.4 −39.2 −116.9 98.1 0.29 92.1 0.43
HN-5 a 339 046 1.5  ±  0.4 −37.5 −135.3 99.8 0.26 95.4 0.38
HN-5 b 275 921 0.0  ±  0.5 −46.6 −161.8 99.8 0.18 98.3 0.26

Abbreviations: PC  =  prostate; MC  =  breast with simultaneously integrated boost; HN  =  head and neck.

5. Discussion

While online transmission detector measurements are performed during plan delivery, the
actual verification is typically done offline. The IQM system is currently the only transmission
detector system enabling real-time intrafraction dose verification.
In this study, the performance of the new transmission detector system was characterized
regarding dosimetric quantities during rotation, accuracy of dose verification, error detection
sensitivity and control point resolved beam monitoring.
In contrast to other transmission detectors (Riley et al 2011, Wong et al 2012, Thoelking
et al 2016), the IQM system provides a 1D spatially sensitive signal for each control point that
does not display spatial dose distribution. This is the key weakness of the IQM system, since
2D dose verification is the most widely accepted method for treatment plan quality assur-
ance. However, the results of this study demonstrates the robust, stable and highly sensitive

7446
Phys. Med. Biol. 62 (2017) 7440 M Pasler et al

Figure 3.  Upper row shows DVH for an exemplary PC (left), MC (middle) and HN
(right) VMAT plan with and without errors (arrow indicates original plan). Second row
shows deviation in (%) between measured reference signal (original plan) and measured
signal of various error-plans:  ×1  ×  2  ±  1 refers to 1 mm shifts of both leaf banks for all
control points toward (−) and away (+) from the central axis of the beam, thus reducing
(−) and increasing (+) field size. MLC shifts that increased the open field area caused
an increase in IQM signal, errors that reduced the open field area also reduced the IQM
signal compared to the signal of the original plan. Detailed data for unilateral MLC
shifts is shown in supplementary figure 2. Abbreviations: PC  =  prostate; MC  =  breast
with simultaneously integrated boost (SIB); HN  =  head and neck; MLC  =  multi leaf
collimator; MU  =  monitor units.

performance of the IQM system as well as good agreement between the calculated and meas-
ured signals meaning it is therefore a powerful tool for beam monitoring.
Recently, Spreeuw et al (2016) have introduced real-time 3D dose verification using portal
imaging. To test their system, two types of serious errors were simulated: (i) MUs were dou-
bled and (ii) MLC and jaws were retracted 10 cm on both sides. Their system detected these
large errors early during delivery. However, error detection sensitivity was not yet tested.
While EPID dosimetry needs no further equipment and additionally accounts for changes in
patient anatomy, full field coverage is not provided and potential image distortions resulting
from mechanical sag during gantry rotation (Rowshanfarzad et al 2012) might increase meas-
urement uncertainty.
Error detection sensitivity tests in this study showed that systematic MLC bank shifts
of 1 mm (signal deviation 5.4  ±  0.9%) and linac output increase of 2% (signal increase
2.1  ±  0.1%) were detected by the cumulative detector output. These results are comparable
to other transmission detector systems: Thoelking et al (2016) reported a 1 mm leaf bank shift
and a wrong linac output of 2% with the Dolphin system; Poppe et al (2010) identified 2 mm
errors for 3 MLC leaf pairs with DAVID; Riley et al (2011) found a 3 mm shift in a single
leaf bank with Delta4AT and Li et al (2016) reported the ability of the Delta4AT system to
quantify MLC position deviations of 0.7 mm during static or 1.2 mm during dynamic IMRT
delivery at static gantry angles. In general, 2D arrays provide spatial information about the
location of errors, hence single MLC malfunction can not only be identified but also localized.
The IQM on the other hand provides a 1D signal that cannot be interpreted any further.

7447
Phys. Med. Biol. 62 (2017) 7440 M Pasler et al

Figure 4.  Correlation of transmission detector signal deviation and DVH difference for
PTV (a) and OARs (c) and (d), mean y-value (e) and passing rate (f) for the 2%2 mm
criterion. (b) Illustrates the cumulative IQM signal for the first 35 control points of a
VMAT plan with  ±1/2 mm MLC error on both banks and the corresponding 1% and
5% deviation corridors, as displayed in real-time during VMAT delivery. Data present
average values and standard deviations (error bars). Abbreviations: PC  =  prostate;
MC  =  breast with simultaneously integrated boost; HN  =  head and neck.

7448
Phys. Med. Biol. 62 (2017) 7440 M Pasler et al

Based on the results of long term signal reproducibility (figure 2(c)), we propose a tol-
erance level for cumulative signal deviation of 1%. All introduced errors (except for uni-
directional MLC shifts) resulted in a signal deviation  >1% and were therefore detectable.
Moreover, IQM signal deviations show a high correlation with DVH deviations (e.g. PTV
Dmean: R2  =  0.97) from reference as well as γ-value evaluation (eg. R2  =  0.86 for 2%/2 mm)
when errors are introduced. However, finding appropriate tolerance levels for halting treat-
ment delivery during VMAT is not trivial and will be the subject of further investigations.
The dosimetric performance of the IQM is found to be very stable in terms of linearity
(R2  =  1), dose rate dependence (<0.4%) up to 460 MU min−1 and long term reproducibility
(<1%). The pre-clinical signal calculation agreed well with measurements (<2.5% devia-
tion), however a larger cohort of VMAT plans is needed to investigate calculation accuracy.
Control point resolved beam monitoring revealed low reproducibility and therefore IQM
segment-by-segment measurements cannot be used for real-time VMAT verification. One rea-
son for that could be the definition of control points for VMAT: In a planning system, control
points are defined as the delivery of a fixed number of MUs with a fixed aperture shape at
a fixed gantry angle (Yu et al 2011). However, the optimized MUs are never intended to be
delivered at the planned gantry angle in VMAT, hence small errors in the beam angle have a
negligible impact on dose deposition. The IQM system on the other hand relies on the gantry
angle to define the boundaries of control points rather than MUs. Considering a VMAT plan at
4° gantry spacing, a 1° gantry angle error would therefore implicate a 25% signal error. Since
gantry overshooting is a known attribute (Pasler et al 2015), large deviations for the segment-
by-segment analysis are to be expected.
Two aspects that need to be addressed when introducing a transmission detector to clinical
routine are the loss of the light field and the additional workload for operating an independent
measurement system. Although the new detector system requires minimal user interaction,
since patient data is loaded via the direct interface to the linac, the system needs to be super-
vised and decisions have to be made about when to halt the linac. Therefore, adequate staff
training and multidisciplinary teamwork is mandatory.
Furthermore, the change in beam characteristics when placing the transmission detector in
the beam path has to be accounted for during treatment planning. Hoffmann et al (2017) report
a beam attenuation of 5.43  ±  0.02%, 4.60  ±  0.02%, and 4.21  ±  0.03% for 6, 10 and 15MV
photon beams. Casar et al (2017) investigated beam hardening effects and the influence of the
detector on surface dose and found a small change in beam quality (0.11–0.53%) and less than
3.3% increase in surface dose for different field sizes at SSD 90 cm.
The new detector system covers risk management at three stages: (i) treatment information
transfer (patient- and plan-ID, gantry- and collimator angle), (ii) dose verification and (iii)
monitoring of treatment delivery. According to the radiotherapy risk profile (WHO 2008), mis-
information or errors in data transfer constitute the greatest bulk of incidents in modern radio-
therapy. All three stages were successfully investigated with the IQM system within this study.
The presented work focuses on VMAT treatment plans only, but static treatment techniques
are equally well supported. In fact, the signal calculation for single beam segments agrees
more accurately with measured signals because no gantry movement is involved. Islam et al
(2009) found an average agreement of 0.5% and 0.9% for individual segments with a maxi-
mum deviation of 3.5% and 5.0% for a prostate and a head and neck IMRT plan, respectively.
Further investigation is needed to evaluate the accuracy of the calculated detector signal for
VMAT deliveries and to find an alternative boundary setting for VMAT control points. Future
work will focus on appropriate tolerance levels for VMAT, IMRT and 3D-CRT as well as the
applicability of the transmission detector system for linac specific VMAT QA and dosimetry
audits.

7449
Phys. Med. Biol. 62 (2017) 7440 M Pasler et al

6. Conclusion

The investigated transmission detector is the first gantry-mounted system to operate in real-
time during patient treatment and monitors the reference versus actual signal. The results of
this study showed excellent long term signal reproducibility and a high correlation with DVH
deviations and γ-evaluation. Therefore, the IQM is suitable for online beam monitoring in
clinical routine.

Acknowledgement

The authors are grateful to iRT for loan of the transmission detector system as well as the fruit-
ful discussions and technical advice. Kilian Michel received a grant from iRT.

References

Casar B et al 2017 Influence of the integral quality monitor transmission detector on high energy photon
beams: a multi-centre study Z. Med. Phys. 27 232–42
Fuangrod T et al 2016 Investigation of a real-time EPID-based patient dose monitoring safety system
using site-specific control limits Radiat. Oncol. 11 106
Hoffmann  D, Chung  E, Hess  C, Stern  R and Benedict  S 2017 Characterization and evaluation of an
integrated quality monitoring system for online quality assurance of external beam radiation
therapy J. Appl. Clin. Med. Phys. 18 40–8
Islam  M  K et  al 2009 An integral quality monitoring system for real-time verification of intensity
modulated radiation therapy Med. Phys. 36 5420–8
Li T, Wu J, Mathen T, Yin F and O’Daniel J 2016 Diode-based transmission detector for IMRT delivery
monitoring: a validation study J. Appl. Clin. Med. Phys. 17 1–10
Mans A et al 2010 3D Dosimetric verification of volumetric-modulated arc therapy by portal dosimetry
Radiother. Oncol. 94 181–7
McNiven  A  L, Sharpe  M  B and Purdie  T  G 2010 A new metric for assessing IMRT modulation
complexity and plan deliverability Med. Phys. 37 505–15
Olaciregui-Ruiz I, Rozendaal R, Mijnheer B, van Herk M and Mans A 2013 Automatic in vivo portal
dosimetry of all treatments Phys. Med. Biol. 58 8253–64
Pasler M et al 2015 Linking log files with dosimetric accuracy—a multi-institutional study on quality
assurance of volumetric modulated arc therapy Radiother. Oncol. 117 407–11
Poppe B, Looe H K, Chofor N, Rühmann A, Harder D and Willborn K C 2010 Clinical performance of
a transmission detector array for the permanent supervision of IMRT deliveries Radiother. Oncol.
95 158–65
Riley S, Matzen T and Carver A 2011 Testing of prototype Delta4AT diode array for in vivo radiotherapy
dosimetry Med. Phys. 28 3534
Rowshanfarzad P, Sabet M, O’Connor D J, McCowan P M, McCurdy B M and Greer P B 2012 Detection
and correction for EPID and gantry sag during arc delivery using cine EPID imaging Med. Phys.
39 623–35
Spezi E, Angelini A L, Romani F and Ferri A 2005 Characterization of a 2D ion chamber array for the
verification of radiotherapy treatments Phys. Med. Biol. 50 3361–73
Spreeuw  H, Rozendaal  R, Olaciregui-Ruiz  I, González  P, Mans  A and Mijnheer  B 2016 Online 3D
EPID-based dose verification: proof of concept Med. Phys. 43 3969–74
Thoelking  J et  al 2016 Patient-specific online dose verification based on transmission detector
measurements Radiother. Oncol. 119 351–6
WHO 2008 Radiotherapy risk profile Technical Manual World Health Organization WHO/IER/
PSP/2008.12
Wong J H D et al 2012 Characterization of a novel two dimensional diode array the magic plate as a
radiation detector for radiation therapy treatment Med. Phys. 39 2544–58
Yu  C  X and Tang  G 2011 Intensity-modulated arc therapy: principles, technologies and clinical
implementation Phys. Med. Biol. 56 R31–54

7450