White Paper October 2013 D913.200.

06/00

Dose reconstruction in the OCTAVIUS 4D phantom and in the patient

without using dose information from the TPS
B. Allgaier, E. Schüle, J. Würfel

PTW-Freiburg Physikalisch-Technische Werkstätten Dr. Pychlau GmbH

Lörracher Straße 7, 79115 Freiburg, Germany

1. Introduction measurements, combined with a dose reconstruction

method that determines dose values in the 3D volume. Such
Intensity modulated radiation therapy (IMRT) and volumetric 2D detectors are available either as planar detector panels
modulated arc therapy (VMAT) are state-of-the-art [7], [8], [9] or as the surface of a cylinder [10]. All these
irradiation techniques for the delivery of highly conformal systems measure time-resolved dose values at a limited
radiation fields to the target volume. These techniques number of positions and reconstruct 3D dose values in the
require complex treatment planning system (TPS) complete volume. While some systems [7], [9] perform this
algorithms as well as sophisticated irradiation methods. As a task as independent measuring tools other systems [8], [10]
result, the use of quality assurance tools for the verification require the dose distribution calculated by the TPS as an
of the planned dose distribution prior to the treatment of the input for the dose reconstruction. Many clinical physicists
patient has become a standard procedure in clinical routine. prefer QA tools to be independent of the object to be tested.
Therefore, the OCTAVIUS 4D was designed as a truly
Early quality assurance (QA) tools for this purpose are two- independent tool for pre-treatment quality assurance. This
dimensional devices based on stationary 2D detector panels paper describes the algorithms on which the OCTAVIUS 4D
[1], [2], [3], [4], [5]. They allow the measurement of one dose reconstruction in the phantom and in the patient are
plane of dose values in order to compare it with the based. In addition, some of the results of the system
corresponding plane calculated by the TPS. Some of these verification are presented.
tools have restrictions regarding the angle of the incident
beam because of the angular dependence of the detector
panel [4], requiring so-called collapsed beam measurements 2. 3D dose reconstruction methods
during which the gantry of the treatment unit is fixed at a
certain angle. Other tools allow composite plan 2.1 TPS dependent methods
measurements by correcting for the directional dependence One example of a QA system for 3D dose measurements is
either by the application of correction factors [1], [5], or by ArcCHECK with 3DVH option [10]. The ArcCHECK
the use of suitably shaped phantoms [3]. detectors are located in a plane that forms the surface of a
cylinder, allowing the radiation beam to hit at least the
Because of the complex volumetric shape of the highly central detectors perpendicularly at any gantry angle. Dose
conformal fields the limitation to one measuring plane is reconstruction in the entire volume of the cylinder is based
considered a disadvantage. Ideally, a full 3D dose matrix is on the modification of the dose matrix calculated by the
desirable. To date the only realistic 3D dosimetry system is TPS. Therefore, before ArcCHECK can determine the
based on gel dosimetry [6], but unfortunately there are volumetric dose grid, the treatment plan including its dose
restrictions in the use of such systems in clinical routine values needs to be entered into the ArcCHECK software.
such as the availability of gels with reproducible features, Then the TPS data are “corrected” by the measured values,
temperature dependence, the change of the irradiated gel and the corrected TPS data are presented as the
volume by time as well as the necessity of relatively ArcCHECK measuring result. This method is called
complex equipment for the evaluation of the irradiated gel “ArcCHECK Planned Dose Perturbation” (ACPDP) [10]. The
samples. scaling factors for the perturbation of the plan are
interpolated from the measured entry and exit values of the
As an alternative to real three-dimensional radiation phantom, resulting in a “morphing” of the dose grid rather
detectors 2D detector panels can be used for the dose than a simple re-normalization. In order to determine the
dose distribution in the patient, the correction factors measuring interval ensures a good correlation between the
obtained for the phantom are simply used for the calculated dose data and the gantry angles. Assuming the currently
patient plan as well, ignoring the patient size and structure highest gantry speed of 360° per minute, the uncertainty of
[10]. In summary, the ArcCHECK method for dose the gantry angle due to the 200 ms measuring interval
reconstruction is based on the calculated treatment plans for arises to ± 0.6°. Combining this value with the inherent
the phantom and for the patient. Without the import of TPS uncertainty of the inclinometer of ± 1° leads to a total
dose data, ArcCHECK is not able to present 3D dose grids. uncertainty of the gantry angle for the 3D dose
Therefore, the ArcCHECK method cannot be considered to reconstruction algorithms of ± 1.2° at a gantry speed of 360°
be independent of the treatment plan that is to be verified. per minute. Radiological tests of the inclinometer’s accuracy
can be found in the literature [7], [12].
In a similar way, the Delta-4 [8] system requires the input of
dose information from the TPS in order to establish a 3D During a typical VMAT measurement over two minutes
dose grid. The Delta-4 modifies the treatment plan OCTAVIUS 4D directly measures more than 400,000 dose
analogously to ArcCHECK. points, distributed over the cylindrical volume of the
phantom.
2.2 Stand-alone methods
Stand-alone QA tools do not use the dose grid calculated by
the TPS in order to determine measured dose distributions
in the phantom or in the patient, they need the TPS dose
grid only for the comparison with the measurements.

Examples for stand-alone systems are COMPASS [9] and

OCTAVIUS 4D [7], [12]. While the COMPASS algorithm is
more complex and constitutes a treatment planning system
on its own [11], the OCTAVIUS 4D algorithm is very straight
forward and requires only few fundamental input data, i.e.
percentage depth dose curves (PDDs) for different field
Fig. 1 The OCTAVIUS 4D system. The detector panel is inserted in
sizes. a cylindrical phantom that rotates synchronously with the gantry.
Detector panels with different spatial resolutions are available. The
The basic version of OCTAVIUS 4D reconstructs dose grids Detector 729 features 10 mm resolution, the central region of the
SRS
in the phantom. The optional “DVH 4D” algorithm allows Detector 1000 features 2.5 mm.
dose volume histograms in the patient to be calculated.

3.2 Commissioning of OCTAVIUS 4D

3. OCTAVIUS 4D dose reconstruction The OCTAVIUS 4D algorithm is based on dose
measurements at a certain depth in the phantom and on
percentage depth dose curves (PDDs) that are used to
3.1 Data acquisition with OCTAVIUS 4D
reconstruct dose values along the ray lines that connect the
The OCTAVIUS 4D phantom (Fig. 1) rotates synchronously
relevant detectors and the focus of the beam. As a
with the gantry, taking time- and gantry angle-resolved dose
consequence, the commissioning of OCTAVIUS 4D is very
measurements. In the course of the measurements the
simple and limited to PDD measurements and to the
complete phantom volume is covered by measuring points.
adjustment of the phantom’s density in the TPS.
Depending on the application, detector panels with different
spatial resolution can be used in the rotational phantom.
For commissioning, PDDs should be measured in a water
The beam incidence is always perpendicular to the surface
phantom at a source-to-surface-distance of 85 cm. PDDs for
of the detector panel, avoiding the need of angular
field sizes of 4 x 4 cm², 10 x 10 cm² and 26 x 26 cm² are
correction factors. The data is collected and evaluated by
mandatory, additional field sizes may slightly improve the
means of the VeriSoft software package that is part of the
accuracy of the reconstruction algorithms.
OCTAVIUS 4D system.

In the TPS, the electron density of the OCTAVIUS 4D

Data is accumulated over measuring intervals of 200 ms.
phantom relative to water should be set to 1.016. If this is
This measuring interval is considered the best compromise
not supported by the TPS, the phantom’s mass density or
between most accurate dose data and optimal correlation
the corresponding Hounsfield Unit used by the TPS is to be
between dose and gantry angle values. Integrating data
adjusted in accordance with the manufacturer’s suggested
over 200 ms results in a high signal-to-noise ratio and dose
procedure [13]. The phantom can be imported into the TPS
values that do not need further smoothing. In this context,
by using the artificial CT file provided with the
the use of ionization chambers instead of silicon diodes is
OCTAVIUS 4D system, or by performing a CT scan without
advantageous as chambers usually have a better signal-to-
the detector panel in place.
noise ratio by design. On the other hand, a 200 ms

PTW D913.200.06/00 2
3.3 Dose distribution in the OCTAVIUS 4D reconstructed this way. The dose reconstruction time
phantom needed by OCTAVIUS 4D is approximately 25 s.
OCTAVIUS 4D and its associated software VeriSoft are
designed to measure and reconstruct a 3D dose grid without 3.4 Dose distribution in the patient
using dose information from the treatment planning system. In order to reconstruct the 3D dose distribution in the
Therefore, the measuring results of the QA test tool are patient, the CT data containing the densities of the
completely independent of the treatment plan. The basics of structures must be imported into VeriSoft. The patient dose
the 3D dose reconstruction algorithm are easy to reconstruction follows the steps described in chapter 3.3,
understand: however with a different procedure in step g). In contrast to
the phantom, the patient is not homogeneous along the ray
a) Convert the PDDs measured in water upon line but features structures with different densities. Also, the
commissioning to PDDs in the OCTAVIUS 4D source-to-surface-distance differs because of the irregular
phantom, using the known relation of the electron shape of the patient contour. This situation is depicted in
densities of water and phantom material Fig. 2.
b) At the current gantry angle (time) consider one
detector of the detector panel (“current detector”)
c) Measure the dose DDet in Gy at this position
d) Construct a ray line through the current detector to the
focus of the beam
e) Determine the current field size from the irradiated
detectors
f) Apply corrections for non-central-axis TPRs according
to [14] in case of beams with flattening filter, or
according to [15] in case of flattening filter free (FFF)
beams
g) Inside the phantom, using the PDD appropriate for the
current field size, reconstruct dose values D(r) in Gy at
the distance r from the current detector along the ray
line according to equation (1)
h) Do this for all detectors of the detector panel
i) Do this for all gantry angles
j) Sort all of the dose values obtained this way into
voxels of 2.5 x 2.5 x 2.5 mm³ (can be changed by the
user) by linear interpolation
k) For the Detector 729, remove a layer of 3 cm in
thickness from the outer shell of the phantom to obtain
a dose grid in a cylinder with 26 cm in diameter and
Fig. 2 OCTAVIUS 4D phantom with ray line (3) through focus and
26 cm in length. For the Detector 1000 SRS the
current detector (2), not to scale. The contour (4) is the patient
reconstructed cylinder is 11 cm in diameter and 11 cm surface from the CT image. The current detector measures the
in length. dose DDet at the water-equivalent depth zPhantom. The algorithm
reconstructs the dose DCT for the current voxel (1) at the water-
The dose D(r) along a ray line is obtained from the dose equivalent depth zCT in the CT image. aDet and aCT are geometrical
DDet(0), measured by the current detector, using the distances from the focus to the current detector and to the current
following relationship voxel, respectively.

PDD(r )
D(r )  DDet (0)  (1) To determine the patient dose at a point on the ray line, a
PDD(0) relationship between the dose measured in the phantom by
the current detector, DDet, and the dose at the point on the
where PDD(r) and PDD(0) are the percentage depth dose ray line in the CT image, DCT, is established. The steps of
values at the distance r from the current detector and at the the algorithm are:
position of the current detector, respectively.
g.1) Convert the PDDs to tissue-phantom-ratios (TPRs)
The outer shell of the phantom is removed from the g.2) Select a point (“current voxel”) on the ray line
reconstructed data as in this region no detectors exist. constructed in d)
Scattered radiation is accounted for by the detectors g.3) Along the ray line, convert the Hounsfield units from
surrounding the current detector. For a typical VMAT plan the CT to electron densities [16]
approximately 1.1 million voxels (dose points) are

PTW D913.200.06/00 3
g.4) Determine the water-equivalent depth of the current
detector in the phantom, zDet, by multiplication of the
geometric depth with the ratio of the electron densities
of the phantom material and water
g.5) Determine the water-equivalent depth of the current
voxel in the patient, zCT, according to equation (3)
g.6) Determine the geometrical distances between focus
and current voxel, aCT, and between focus and current
detector, aDet
g.7) Calculate the dose in the patient DCT at the current
voxel by application of equation (2)
g.8) Do this for all voxels along the ray line inside the
patient’s contour.
Fig. 3 Dose volume histograms in the patient’s geometry
determined by OCTAVIUS 4D (dashed lines) and by the TPS (solid
The reconstruction of the dose in the patient (or in the CT
lines)
image) along a ray line through the current detector and the
focus is based on equation (2). The meaning of the symbols
is explained above.
Mostly, dose distributions in the patient are visualized by
means of dose volume histograms (DVHs). Usually, DVHs
2
TPR ( zCT )  a Det  are presented not for the complete patient but for different
DCT  DDet    (2)
TPR ( z Det )  aCT
structures such as organs at risk or target volumes. In order
 to determine DVHs the VeriSoft software must import the
geometry of the patient structures from the TPS. The DVH
The first term in equation (2) accounts for different algorithm uses the reconstructed 3D dose distribution in the
thicknesses of overlaying material in front of the current patient and determines the histograms from the dose values
detector in the phantom and in front of the current voxel in inside the structures. VeriSoft only considers structures that
the patient (or CT). The water-equivalent depth zCT of the are completely inside the OCTAVIUS 4D measuring volume.
current voxel in the patient is obtained from the geometrical If the TPS allows DVH curves to be exported, they can be
geom
depth z CT and the electron densities of the materials compared with the DVH curves determined by
OCTAVIUS 4D as depicted in Fig. 3.
involved:
n

 CT , i
4. Reconstruction accuracy
zCT  z geom
 i 1
(3)
n  Water
CT

4.1 Test methods

The absolute accuracy of systems like OCTAVIUS 4D is
where n is the number of voxels along the ray line from the
hard to determine for all clinical situations. Besides other
current voxel to the surface of the patient, Water is the parameters, the accuracy depends on the complexity of the
electron density of water, and  CT , i is the electron density
treatment plan, on the presence of dose gradients and
regions of low density, and on the spatial resolution of the
of voxel i (i = 1, 2, …, n). detector panel used for the measurements. For this paper,
two methods were chosen to estimate the accuracy of the
The second term in equation (2) accounts for the different measured dose distribution in the phantom: comparison of
distances to the focus by applying the inverse square law. static field results with TPS calculations (chapter 4.2), and
comparison of complex clinical IMRT and VMAT plans with
The TPRs in equation (2) are transformed from the the results of a treatment planning system, as well as with
measured PDDs by application of the inverse square law measurements obtained by a stationary OCTAVIUS II
according to BJR 11 [17]. system (chapter 4.3).

This algorithm is not able to account for the change of The accuracy of the DVHs was also tested employing two
scattered radiation as the position on the ray line passes different methods. First, the reconstructed maximum dose in
structures with different density. the patient was compared with the TPS dose calculation.
Second, a full-volume gamma analysis on the reconstructed
The time needed by OCTAVIUS 4D to reconstruct the 3D dose volume was performed by comparison with the
dose distribution in the patient is typically in the order of two treatment plan (chapter 4.4).
minutes.

PTW D913.200.06/00 4
Before the measurements and the data analysis the
OCTAVIUS 4D system was correctly commissioned as
described in chapter 3.2. Special attention was given to the
correct adjustment of the phantom’s electron density in the
treatment planning system.

Unless otherwise noted, all comparisons of dose

distributions were made with VeriSoft 6.0 in form of a 3D
gamma analysis with a local gamma criterion of 3% / 3 mm,
suppression of dose values below 10% of the maximum
dose, and with the “2nd and 3rd pass” option activated [18]. In
the following, the relative number of voxels that pass this
gamma criterion is referred to as “pass rate” and serves as
a measure for the goodness of the reconstructed dose
distribution.
Fig. 4 Static field 10 x 10 cm², coronal plane, isocentric cross
The measurements were made on a Varian Clinac iX linear profile: OCTAVIUS 4D with Detector 729 (spatial resolution 10 mm,
accelerator and compared with the treatment plans created green curve) versus TPS (yellow curve). The pass rate in the
by the Varian Eclipse V10 treatment planning system. A complete plane was 83.3%.
comprehensive verification of the OCTAVIUS 4D system
can also be found in the literature [7], [12].

4.2 Comparison with static fields

A simple way to get an idea of the accuracy of the
implemented algorithm is the application of a static field to
the OCTAVIUS 4D, and the comparison of the
reconstructed cross profiles and the reconstructed depth
profile with the corresponding profiles obtained from the
TPS. It can be assumed that the accuracy of the TPS for
such a simple geometry is good enough to serve as
reference. Figures 4 - 6 show the results of a static 10 x
10 cm² beam evaluated with OCTAVIUS 4D. It can be seen
from Fig. 4 that with the standard detector panel
(Detector 729) the profiles show deviations in the penumbra
regions because of the panel’s spatial resolution of 10 mm.
These deviations, however, become less important as the Fig. 5 Static field, 10 x 10 cm², coronal plane, isocentric cross
SRS
beam rotates, averaging over many measuring angles and profile: OCTAVIUS 4D with Detector 1000 (spatial resolution
2.5 mm, green curve) versus TPS (yellow curve). The pass rate in
smearing out the resolution effect. This was demonstrated
the complete plane was 100% (95.5% at 1% / 1 mm).
by the measurement of a full arc irradiation using the same
field, resulting in a cylindrical dose distribution that matched
the distribution calculated by the TPS with a pass rate of
99.8% as compared to 83.3% for the static field.

Using a detector panel with higher spatial resolution such as

the OCTAVIUS Detector 1000 SRS almost eliminates the
resolution effect even for static beams. The spatial
resolution in the central area (5.5 x 5.5 cm²) of this detector
panel is 2.5 mm, and the pass rate obtained was 100%. As
a result, the measured profiles are in very good agreement
with the TPS profiles (Fig. 5).

Fig. 6 depicts a 6 MV depth dose curve obtained with the

OCTAVIUS 4D with Detector 729, applying a static 10 x
10 cm² field. The conformance with the TPS is very good.
Fig. 6 Static field, 10 x 10 cm², transverse plane: Central axis
depth dose distribution obtained by OCTAVIUS 4D with
Detector 729 (green curve), compared with TPS (yellow curve). The
pass rate in the complete plane was 88.3%.

PTW D913.200.06/00 5
4.3 Comparison with a stationary detector panel A similar comparison between OCTAVIUS II and
and with TPS results OCTAVIUS 4D was made by the delivery of a RapidArc lung
A more realistic test is the comparison of OCTAVIUS 4D plan with nodes. Here the pass rates were 100% for
with a system using a stationary detector panel by applying OCTAVIUS II and 99.6% for OCTAVIUS 4D.
an IMRT plan or a rotational plan. For this purpose, an
OCTAVIUS II system, consisting of a stationary octagonal Finally, a clinical prostate plan was measured with
phantom and a Detector 729, was irradiated with an IMRT OCTAVIUS 4D using detector panels with different spatial
Head & Neck plan consisting of 11 fields. The coronal plane resolution. Fig. 8 shows the results for the Detector 729 with
defined by the detector panel was evaluated with VeriSoft. 10 mm resolution and for the Detector 1000 SRS with 2.5 mm
Then the same irradiation plan was applied to an resolution. At 3% / 3 mm both pass rates were 100%,
OCTAVIUS 4D system, and the same coronal plane was therefore the criterion was changed to 2% / 2 mm. The pass
extracted from the 3D dose grid. Fig. 7 shows the rates were then 95% and 98.5%, respectively.
comparison of both measurements with the TPS. The
OCTAVIUS II achieved a pass rate of 99.8%, while the pass
rate of the OCTAVIUS 4D was 99.2%.

Fig. 8 Effect of spatial resolution of the detector panel. The upper

graph shows a profile in the coronal plane obtained with the
Detector 729, the lower graph depicts the equivalent profile
SRS
obtained with the Detector 1000 . Both detector panels were
used with the OCTAVIUS 4D system. The corresponding TPS
profiles are overlaid in yellow color. The pass rates in the complete
plane for a gamma criterion of 2% / 2 mm were 95% and 98.5%,
respectively.
Fig. 7 Comparison of dose values in the coronal plane obtained by
a stationary OCTAVIUS II system (upper graph) and a rotational
OCTAVIUS 4D system (lower graph). The irradiated plan was a 4.4 Test of DVH accuracy
Head & Neck IMRT plan with 11 fields. The pass rates were 99.8% The accuracy of the dose volume histograms determined by
and 99.2%, respectively. OCTAVIUS 4D has been evaluated by comparing them with
DVHs exported from the treatment planning system. All
measurements were made using the Detector 729. For a set

PTW D913.200.06/00 6
of clinical treatment plans the maximum doses were additional data consist of PDDs at different field sizes and
determined from the reconstructed plan and from the TPS. the phantom’s electron density. For the dose distribution in
The differences between reconstruction and TPS were the patient, geometric information on the patient’s structures
noted and their mean values calculated. The uncertainty of and their densities are needed in addition. No dose
the maximum DVH dose measured with OCTAVIUS 4D information from the TPS is used to establish the measured
was found to be ± 2% for relatively homogenous plans, and dose distributions, making the OCTAVIUS 4D system a real
± 6% for lung plans. These uncertainties are approximately independent QA tool.
twice as high as the uncertainty of the DVHs calculated by
the TPS. The biggest discrepancies between OCTAVIUS 4D and
TPS data is found for cross profiles measured with the
For a deeper insight into the accuracy of the reconstructed Detector 729 in static fields (Fig. 4). This is the result of the
3D dose grid in the patient, the commercial version of relatively coarse spatial resolution of the detector panel. The
VeriSoft was modified to allow the comparison of individual detectors are 0.5 x 0.5 x 0.5 cm³ in size and
reconstructed dose planes in the patient with those from the separated by 10 mm center to center. The fact that the
TPS. Fig. 9 shows the result of such a gamma analysis in a phantom normally rotates with the beam smears out this
patient’s plane. resolution effect to a big degree, increasing the pass rate in
the evaluated plane from 83.3% to 99.8%. Also, the use of a
detector panel with a higher spatial resolution improves the
results (Fig. 5).

The measurement of a static depth dose curve shows very

good agreement with the TPS (Fig. 6). Nevertheless, the
pass rate of 88.3% for the evaluated transverse plane is
relatively low. This is obvious as the gamma index analysis
covers not only the central axis shown in Fig. 6 but also the
complete field including the penumbra regions. Again, the
limited spatial resolution of the Detector 729 is responsible
for this effect.

Fig. 7 compares dose distributions reconstructed by

OCTAVIUS 4D and measured directly by a stationary
Fig. 9 Dose plane in the patient’s geometry, comparison between detector panel, avoiding the need of dose reconstruction
OCTAVIUS 4D and TPS. The red failed points at the bottom reflect
algorithms. While the stationary OCTAVIUS II system only
the fact that the reconstruction algorithm included the air between
evaluates the gamma values for each detector on the panel,
the patient and the couch whereas the TPS did not. For this gamma
analysis, the commercial version of the VeriSoft program was the rotating OCTAVIUS 4D system calculates gamma
modified and the results compared with a Pinnacle treatment values for all reconstructed voxels in the complete volume.
planning system. The pass rate at 2% / 2 mm was 93.3%. The reconstruction of voxels is based on a linear
interpolation between detectors which leads to small
uncertainties. As a result, the pass rate of the
In order to estimate the dose errors in the reconstructed OCTAVIUS 4D with Detector 729 is slightly lower than that
patient, a series of gamma analyses in the complete volume of the OCTAVIUS II, especially in regions of steep dose
was made by varying the dose criterion while keeping the gradients.
distance-to-agreement criterion constant at 2 mm. From the
resulting relationship between dose criterion and pass rate it A comparison of OCTAVIUS 4D with head & neck plans of
could be seen that a pass rate of one standard deviation 50 patients has been reported in the literature [12]. The
(k = 1) corresponded to a dose error of approximately ± 2% average pass rate found was (99.2 ± 0.8)%. These
for relatively homogeneous plans and ± 5.5% for lung plans. measurements included high-dose rate, flattening filter free
This is in reasonable agreement with the differences found (FFF) beams. This indicates that the OCTAVIUS 4D
by comparing the maximum doses as described above. reconstruction algorithm works for both flattened and
unflattened beams.

5 Discussion The data shown in Fig. 8 clearly demonstrate the fact that
the pass rate for clinical treatment plans raises as the
The algorithms used by OCTAVIUS 4D are straightforward, spatial resolution of the detector panel is increased.
using only the gantry-resolved measured dose matrices, the
associated gantry angles, and a very limited set of additional Although the problem is much reduced by the rotation of the
data provided by the user upon commissioning of the phantom, the spatial resolution of 10 mm for the
system. For the dose distribution in the phantom the Detector 729 and the associated linear interpolation

PTW D913.200.06/00 7
algorithm result in small, but detectable deficiencies in the [3] Ann Van Esch et al., “On-line quality assurance of rotational
radiotherapy treatment delivery by means of a 2D ion chamber
gamma analysis. In this context, however, it is worth noting
array and the Octavius phantom”, Med. Phys. 34 (10), October
that comparable devices are having similar restrictions. The 2007, 3825 - 3837
Delta-4, for instance, also uses a linear interpolation
[4] Paul A Jursinic et al., “MapCHECK used for rotational IMRT
algorithm [8], and the ArcCHECK requires agreement
measurements: Step-and-shot, Tomotherapy, RapidArc”, Med.
between measurements and treatment plan within Phys. 37, (6), June 2010, 2837 - 2846
approximately 10% in order to be able to reliably correct the
[5] M Stasi et al., “D-IMRT verification with a 2D pixel ionization
calculated patient dose distribution [10]. For the COMPASS chamber: dosimetric and clinical results in head and neck cancer”,
system dose errors of up to 17% have been reported [9]. Phys. Med. Biol. 50 (2005) 4681 - 4694

[6] Yoichi Watanabe et al., “3D evaluation of 3DVH program using

Finally, the treatment planning systems themselves are not BANG3 polymer gel dosimeter”, Med. Phys. 40 (8), August 2013,
free of errors. Errors in the range 1% - 3% have been 082101-1 - 082101-11
reported [19], [20], [21]. In the presence of air volumes
[7] Sotirios Stathakis et al., “Characterization of a novel 2D array
(lung) the inaccuracies of the TPS calculations can be even dosimeter for patient-specific quality assurance with volumetric arc
higher by a factor of 3 - 5 [19], [20], [21]. therapy”, Med. Phys. 40 (7), July 2013, 071731-1 - 071731-9

[8] R Sadagopan et al., “Characterization and clinical evaluation of

The uncertainty in the determination of DVHs in the patient a novel IMRT quality assurance system”, Journal of Applied Clinical
originates from measurement inaccuracies and inaccuracies Medical Physics, Vol. 10, No. 2, Spring 2009, 104 - 119
of the reconstruction algorithm used by OCTAVIUS 4D. The
[9] J Godart et al., “Reconstruction of high-resolution 3D dose from
more homogeneous the patient structure of interest is, the matrix measurements: error detection capability of the COMPASS
better the accuracy of the reconstructed DVH. The correction kernel method”, Phys. Med. Bol. 56 (2011), 5029 - 5043
uncertainty of the reconstructed maximum doses of the
[10] Benjamin E. Nelms et al., ”VMAT QA: Measurement-guided 4D
DVHs ranges from ± 2% for relatively homogeneous plans dose reconstruction on a patient”, Med. Phys. 39 (7), July 2012,
to ± 6% for lung plans. This seems to be very adequate 4228 - 4238
considering the fact that no information from the TPS about
[11] R Boggula et al., “Experimental verification of a commercial 3D
the dose distribution is required by the reconstruction dose verification system for intensity-modulated arc therapies”,
algorithm. Phys. Med. Biol. 55 (2010), 5619

[12] Conor K. McGarry et al., “Octavius 4D characterization for

flattened and flattening filter free rotational deliveries”, Med. Phys.
6 Conclusions 40 (9), September 2013, 091707-1 - 091707-11

[13] PTW-Freiburg, Technical Note, “Parameter Settings in VeriSoft

The OCTAVIUS 4D algorithms for the reconstruction of when relative Electron Density cannot be set in the TPS”,
D655.200.07
three-dimensional dose distributions in the phantom and in
the patient have been described. The algorithms are [14] R. C. Taylor et al., ”A generic off-axis correction for linac
completely independent of dose information calculated by photon beam dosimetry”, Med. Phys. 25 (5), May 1998, 662 - 667
the treatment planning system, and they need only very few [15] Dietmar Georg and Gabriele Kragl, „Photon beam quality
basic data for the commissioning. Various validation variations of a flattening filter free linear accelerator“, Med. Phys. 37
procedures and results have been presented, proving the (1), January 2010, 49 - 53
suitability of OCTAVIUS 4D for clinical IMRT and VMAT pre- [16] S J Thomas, “Relative electron density calibration of CT
treatment QA. Validity of the system for use with FFF beams scanners for radiotherapy treatment planning”, The British Journal
has been reported in the literature [12]. The dose accuracy of Radiology, 72 (1999), 781 - 768

of dose volume histograms in the patient’s geometry has [17] “Central Axis Depth Dose Data for Use in Radiotherapy”,
been estimated to vary from ± 2% to ± 6% depending on British Journal of Radiology, Supplement 11, London 1972
the magnitude of the density variations in the patient. The
[18] Tom Depuydt et al., ”A quantitative evaluation of IMRT dose
accuracy of the system can be improved by the use of distributions: refinement and clinical assessment of the gamma
detector panels with higher spatial resolution such as the evaluation”, Radiotherapy and Oncology 62 (2002) 309 - 319
SRS
Detector 1000 . [19] I. Ali et al., “Quantitative assessment of the accuracy of dose
calculation using pencil beam and Monte Carlo algorithms and
requirements for clinical quality assurance”, Medical Dosimetry, in
press, 2013

References [20] Tao Han et al., “Dosimetric comparison of Acuros XB

deterministic radiation transport method with Monte Carlo and
[1] E Spezi et al., “Characterization of a 2D ion chamber array for model-based convolution methods in heterogeneous media”, Med.
the verification of radiotherapy treatments”, Phys. Med. Biol. 50 Phys. 38 (5), May 2011, 2651 - 2664
(2005) 3361 - 3373
[21] M. W. K. Kan et al., “Experimental verification of the Acuros
[2] S. Cilla et al., “Portal dose measurements by a 2D array”, XB and AAA dose calculation adjacent to heterogeneous media for
Physica Medica (2007) 23, 25 - 32 IMRT and RapidArc of nasopharygeal carcinoma”, Med. Phys. 40
(3), March 2013, 031714-1 - 031714-19

PTW D913.200.06/00 8