12 / 146
The purpose of this study was to compare 2D and 4D pre-treatment ver-
ification for 10 patients, focusing on the linac couch attenuation, in particular
for posterior oblique treatment fields.
Material and methods:
The ion chamber matrix PTW 2DARRAY was po-
sitioned between a RW3 phantom and inside the OCTAVIUS4D equipped
with an integrated rotation unit for 2D and 4D verification respectively.
IMRT step and shoot treatments were planned with TPS Oncentra MasterPlan
and delivered with linac Elekta Synergy Platform. 2D pre-treatment veri-
fications were performed setting all fields at gantry 0°. The field to field
gamma-index analysis was performed with Verisoft software.
Results:
The calculated and measured data were compared using DTA 3 mm,
DA 3%. The results show a degradation of field to field comparison for
posterior-oblique beams, with a consequent reduction of the delivered dose
with respect to the planned one. This is due to the linac carbon fiber couch
attenuation. The attenuation factor obtained was on the order of [2.2–
4.5]% at gantry angle range [130°–230°]. To solve this question, a virtual
couch of the same shape and composition of the Elekta Synergy one was
created in the TPS contouring workspace, positioned at the bottom of a cy-
lindrical virtual phantom and it was imposed a density of 1750 kg/m
3
. The
passed point’s gamma-index analysis improved from 64.5% to 94.5%.
Conclusions:
From the data analysis it was shown that the couch atten-
uation is a gantry angle function; therefore, a virtual couch was developed
in the TPS. With the aim of taking into account the couch attenuation factor
during the IMRT planning, the application of the virtual couch will be ex-
tended to each patient’s simulation CT.
http://dx.doi.org/10.1016/j.ejmp.2016.01.025A.22
VALIDATION OF A DEFORMABLE IMAGE REGISTRATION ALGORITHM FOR
RADIOTHERAPY APPLICATIONS
A. Bruno
* , a ,C. Gasperi
a ,A. Rampini
b ,G. Bell
i a .a
U.O.C. Fisica Sanitaria, AUSL8
Arezzo, Arezzo, Italy;
b
U.O.C. Radioterapia, AUSL8 Arezzo, Arezzo, Italy
Introduction:
The goal of this study was to validate the accuracy of a de-
formable image registration (DIR) algorithm provided by the commercial
software MIM Maestro used as a contouring tool in the planning of radi-
ation therapy treatments. This study focused on the evaluation of DIR for
prostate imaging.
Materials and methods:
Evaluation was performed first on synthetic images
simulating a CT scan of male pelvis by providing variable bladder and rectum
filling. Subsequently, tests were performed on CT and MR scans of a home-
made phantom simulating different fillings of bladder and displacements
of prostate. Finally, the study was addressed on clinical images (CT-CT, CT-
MR). Ten patients treated for cancer prostate were selected. For all steps,
DIR was evaluated first by visual assessment, then using similarity mea-
sures (SMs) and eventually inspecting the properties of the registration.
Results:
In synthetic images, the main inaccuracy was found for rectum
partially filled showing geometric distortions that also included the sur-
rounding region. Non-physical deformations of bladder resulted for bladder
fillings less than about 50%. The study on phantom showed how to opti-
mize the use of the module RegRefine for CT-MR registration. Applying
RegRefine on only areas of the bladder and prostate improved Dice simi-
larity index (DSI) about 11% for both phantom and clinical CT-MR images.
Even for clinical images, the critical issue regarded the different levels of
filling of the rectum. The application of RegRefine to the registration of clin-
ical images CT-CT has provided an improvement of DSI for the target of about
3%.
Conclusions:
The accuracy showed by MIM DIR algorithm allows appli-
cations for target contouring. DIR evaluation with SM is necessary to
overcome the limits of visual assessment; however, in clinical practice, DIR
results have to be always validated by clinicians. In the future, evaluation
may be performed using other SM, like Jacobian of the transformation
matrix.
http://dx.doi.org/10.1016/j.ejmp.2016.01.026A.23
ELEKTA ONCENTRA 4.3 AND MONACO 5.0 COMPARISON FOR 3DCRT DOSE
CALCULATION ALGORITHMS
C. Cadioli
* , a ,M.G. Brambilla
a ,A.F. Monti
a ,C. Carbonin
i a ,C. De Mattia
a ,M.B. Ferrar
i a ,D. Zanni
a ,G. Albert
a b ,A. Torresin
a .a
Ospedale Niguarda Ca’
Granda, Milano, Italy;
b
Elekta S.p.A., Agrate Brianza, Italy
Introduction:
Accurate dose tests must be performed before using a treat-
ment planning system (TPS) in clinical practice. Measured and calculated
dose distributions must be compared in various irradiation conditions, and
a huge amount of time is needed. In this work, we evaluated the differ-
ences between 3DCRT calculated dose distributions in the migration between
two TPSs produced by the same company.
Material and methods:
In our centre, the migration from Oncentra ver.
4.3 (Elekta, SWE) to Monaco ver. 5.0 (Elekta, SWE) was carried out. The two
TPSs use the same 3DCRT dose calculation algorithm. The kernels for 3 dif-
ferent photon energies produced by a Synergy (Elekta, UK) with an 80 leaves
MLC were processed and installed on the Monaco console.
The dose distributions calculated by the systems were analysed in terms
of depth doses, profiles at various depths and absolute dose. The compar-
ison between calculation and measurement was performed according to
ESTRO booklet 7 criteria. For relative data, the reference analysis param-
eter was the gamma index confidence limit. The tolerances in gamma index
variables for global and local tests were changed according to the irradi-
ation geometry complexity and a maximum dose threshold of 7–10% was
used. A specific analysis software provided by Elekta Support was used for
the comparisons.
For absolute doses, the reference analysis parameter was the percentage
difference between measured and calculated values (acceptance criteria from
2% to 3% depending on complexity).
Results:
No significant differences between the two TPS calculated doses
were found, except for negligible variations in field shape (around 0.5 mm).
Conclusion:
If an Oncentra accurate commissioning is present, a reduced
number of comparison tests, involving each implemented energy and ra-
diation unit, could be used with Monaco. Our results refer to Oncentra ver.
4.3 and the present considerations should not be adopted for previous ver-
sions without any specific check.
http://dx.doi.org/10.1016/j.ejmp.2016.01.027A.24
RE-IRRADIATION (RE-RT) OF HEAD AND NECK CANCER (HNC): DOSE
ACCUMULATION USING DEFORMABLE IMAGE REGISTRATION (DIR) AND
COMPARISON WITH RIGID REGISTRATION (RIR)
E. Cagni
* , a ,A. Bott
i a ,R. Micera
b ,L. Orsingher
a ,M. Orland
i a ,N. Simoni
a ,C. Iotti
b ,M. Iori
a .a
Fisica Medica, ASMN-IRCCS, Reggio Emilia, Italy;
b
Radioterapia, ASMN-IRCCS, Reggio Emilia, Italy
Introduction
: Re-RT is considered an option for recurrent HNC. In this sce-
nario, it is important to accurately estimate the initial treatment dose
received by the critical organs (OARs). Deformable image registration (DIR)
could improve this accuracy.
The aim of the study was to evaluate the use of DIR to account for initial
treatment. The work is focused on the DIR accuracy estimation and on the
comparison between RIR and DIR in terms of spatial and dosimetric
differences.
Material and methods:
Firstly, the accuracy of DIR algorithm was as-
sessed for 5 HN re-RT patients using distance propagation and Dice similarity
index (DSI). On the same patient group, a validation procedure, based on
inverse consistency (IC) method (Garcìa-Molla, Physica Medica 2015), was
performed.
Secondly, DIR and RIR were retrospectively applied to 10 HN re-RT pa-
tients. Several OARs were automatically transposed from CT1 to CT2 and
compared with manually delineated structure set by means of Hausdorff
distance (HD) and DSI. The dose registration was performed with both
methods and cumulative dose distribution was evaluated in terms of isodose
distances (HD95) and dose-volume histograms (DVH).
Results:
The accuracy of DIR resulted in 2.0
±
1 mm, with a mean DSI of
0.8
±
0.1 with respect to the reference structures. The IC method showed
DSI of 0.9
±
0.0 and 0.9
±
0.1 for contours and isodoses respectively. Overall,
DIR matched manual contours closely than RIR with an improvement of
e7
Abstracts/Physica Medica 32 (2016) e1–e70