831 / 1096
S815
ESTRO 36
_______________________________________________________________________________________________
ensured for the scattering and to mimic the TPS treatment
planning. The result shows that, the total neutron dose
increases as the field size increases from 10 x 10 cm
2
to 20
x 20 cm
2
and for 40 x 40 cm
2
. The photo neutron
measurements using activation foils for Omni wedged
fields in Elekta LINAC is uniquely studied. The irradiation
time of about 20 min were taken to deliver 50 Gy at Dmax
with the dose rate of 640 Mu/min. Wedged fields were
defined for a field size of 30 x 30 cm
2
and the wedge used
for each set of measurements are 15°, 30° and 60°. The
fast neutron dose decreases and thermal neutron dose
increases with wedge angles from 15°, 30° and 60°. Open
beam gives the highest fast neutron dose and the lowest
thermal neutron dose.
Conclusion
Insensitivity nature of activation foils for gamma/photons
and the possibility of absolute measurements using the
primary quantity of nuclear reaction cross-section makes
activation foil best suited for photon induced neutron
measurement. The present results indicate that the total
neutron dose represents a small contribution to the
therapeutic photon dose, meaning that it is much smaller
than 1% of the photon dose delivered to the patient.
However, the amount of this extra dose in the vicinity of
the patient position cannot be neglected in view of
radiological protection assessment related to the patients.
Electronic Poster: Physics track: Treatment plan
optimatisation: algorithms
EP-1519 Implementation of a hybrid superfast Monte
Carlo-Pencil beam dose optimizer for proton therapy
A.M. Barragán Montero
1
, K. Souris
1
, D. Sánchez-
Parcerisa
2
, A. Carabe-Fernández
3
, J.A. Lee
1
, E. Sterpin
1,4
1
Université Catholique de Louvain- Institute of
Experimental & Clinical Research, Molecular Imaging-
Radiotherapy and Oncology MIRO, Brussels, Belgium
2
Universidad Complutense de Madrid, Departamento de
Física Atómica- Molecular y Nuclear, Madrid, Spain
3
Hospital of the University of Pennsylvania, Department
of Radiation Oncology, Philadelphia, USA
4
KU Leuven - University of Leuven, Department of
Oncology, Leuven, Belgium
Purpose or Objective
Monte Carlo (MC) dose calculation plays an important role
in treatment planning for proton therapy due to the
limited accuracy of analytical algorithms, especially in
very heterogeneous tumor sites. The new dedicated MC
engines for pencil beam scanning (PBS) achieve reduced
computation times for a single dose calculation. However,
computing spot-per-spot doses is still very time-
consuming, since typically 10000 to 20000 spots are
needed. The presented strategy combines the speed of
analytical algorithms and the accuracy of MC to get the
best outcome for PBS treatment planning in a reasonable
amount of time for clinical practice.
Material and Methods
An in-house treatment planning system was used to create
the plans. The optimizer combines the analytical pencil
beam (PB) algorithm in
FoCa
(Sánchez-Parcerisa et al.
Phys Med Biol 2014) and the super-fast Monte Carlo engine
MCsquare
(Souris et al. Med Phys 2016) able to compute a
final dose in less than 1 minute.
The hybrid optimization strategy calculates the optimal
spot weights (
w
) using the analytical beamlets matrix (
P
PB
)
and a correction term
C
. After a first optimization where
C
= 0, the method alternates optimization of
w
using
P
PB
with updates of
C = D
MC
–
D
PB
, where
D
MC
results from a
regular MC computation (using 10
8
protons to ensure good
statistical accuracy) and
D
PB
= P
PB
* w
. Updates of C can be
triggered as often as necessary by running the MC engine
with the last corrected values of
w
as input.
The performance of the method is illustrated on two
extreme cases: prostate (relatively easy case) and lung
(considered to be complex due to the high heterogeneity).
For simplicity, we created PTV-based plans but the
findings can be equally applied to robust optimized plans.
Results
For the prostate case, the recomputed MC dose after
initial optimization (
C
=0, before correction) revealed a
decreased target coverage (D95=90% of the prescribed
dose, D
p
) that improved significantly after just one
correction (D95
corrected
=97%D
p
).
For the lung case, the difference between MC and PB doses
before correction was very large: D95=63%D
p
and
D5=137%D
p
. But still the hybrid strategy was able to
partially improve target coverage (D95
corrected
= 84%D
p
) as
well as reducing overdose (D5
corrected
= 111%D
p
), after two
updates of C.
In both cases, further corrections did not lead to better
results.
The results proved that the hybrid method allows us to
improve dose accuracy even for very complicated cases as
lung tumors. However, the success of the correction is
limited by the order of magnitude of the term C, i.e, very
large difference between MC and PB doses are only
partially corrected.