816 / 1096
S800
ESTRO 36
_______________________________________________________________________________________________
Material and Methods
In the treatment prostate plan with VMAT high-
fractionated (2x7.5Gy), FFF technique the errors of dose
(differences ±1%; 2%; 3%; 5% 7%, 10%), collimator angle
(rotations in both directions: 0.5; 1.0; 1.5; 2.0; 2.5; 3.0)
and MLC shifts were introduced. For each modified plan,
the pre-treatment verification plan was created and
measured with 2D-arrays: 729 and SRS 1000 with rotational
phantom Octavius® 4D and Verisoft 6.1 software with DVH
option (PTW, Freiburg, Germany). Measured (with errors)
and calculated (reference plan) dose distributions were
analyzed with 3D gamma evaluation method for various
tolerance parameters DTA [mm] and DD [%] 1.0; 1.5; 2.0;
2.5; 3.0, by global and local dose methods with a 5%
threshold. To detect errors, the achieved score should be
less than the assumed tolerance of 95%. Additional the
DVHs from error-induced and reference plan were
analyzed for CTV D
50
, D
98,
D
2
, and D
25
, D
50
for OARs.
Results
For 12 error-induced plan with dose discrepancies, proper
detection for 729 and SRS 1000 were obtained as follows:
3/12 and 6/12 (G3%/3mm); 8/12 and 6/12 (L3%/3mm);
8/12 and 7/12 (G2%/2mm); 8/12 and 8/12 (L2%/2mm).
The rotations of collimator were detected >3° for 729 and
>2° for SRS 1000. The MLC errors were discovered for plans
with 1 leaf (MLC1) and 1 pair of leaves (MLC2) blocked, for
all leaves shifted about 0.05cm (MLC3) misalignment
weren’t indicated so obvious. The clinical relevance of
plan with MLC errors and chosen discrepancies for
collimator rotation (3°) and dose differences (+5%) were
presented in the table 1.
Conclusion
To more sophisticated analysis the gamma criteria should
be less than 3%/3mm or/and local dose method should be
used. The resolution of used detector is crucial and should
be high for better interpretation of results. Gamma
method presents some statistic data, for scrutiny analysis
the clinical interpretation should be assessed.
EP-1493 Machine record parameters or Epid based
data for ART QA. A comparison of two scenarios.
P. Haering
1
, C. Lang
1
, M. Splinter
1
1
DKFZ, E040, heidelberg, Germany
Purpose or Objective
Using machine record files and Epid based dosimetry is
popular for machine and patient related QA, as this may
also work for adaptive treatment approaches. The
Siemens Artiste treatment machine used here, allows a
comparison of both methods in one session. Exit images
and all relevant machine parameters are included in the
image header collected during treatment. Here we
present results of a comparison between QA dose
recalculations based on the two sources, exit images and
machine recorded parameters.
Material and Methods
A software tool was developed that allows for the
extraction of the relevant parameters (MLC-positions, MU,
etc.) from the machine records as well as from the Epid
measured exit fluencies. While machine data had to
undergo a reformat to be used for recalculation, the exit
fluencies need more attention. Here both, the delivered
fluence as well as the absorption in the
patient do play a
role. Therefore both have to be separated to receive
reliable MLC positions. The algorithm used first generates
an image containing only absorption information for the
beam using this to remove this influence on the MLC
positions. MUs were used from the parameter file, as the
fluence uncertainties on the EPID images have shown to
be to large to be used for that purpose. The extracted
parameters are then inserted in a newly generated Dicom
RT-Plan file that then can be used in the treatment
planning system (here Raystation, Raysearch) to
recalculate the dose. Dose distributions (Epid based,
parameter file based and originally planned) are then
compared.
Results
Measuring exit doses with the EPID was a simple task and
could be done for all coplanar field sets. The software tool
made it simple to extract all the needed parameter from
the files and images resulting in 2 new Dicom plan files.
Dose recalculation was done by just importing the new
plan files to Raystation. Comparing the original dose
distribution to the machine file based one showed almost
no difference at all (< 0.7%), as MU and leaf position
differences where quite small. This might also be
grounded in the used calculation grid of 2mm size. MLC
positions derived from EPID images show much larger
differences. Here detection uncertainties, EPID
positioning and the resulting image resolution of 0.3mm
do play a major role. This resulted in in noticeable
differences in the dose gradients regions. Absolute dose
differences where below 1.5%.
Conclusion
Recalculating doses based on EPID and machine based
parameters is a possible way for QA in an adaptive
treatment approach. As QA parameters are taken from
information that is given anyway or that can be easily
generated, it does not complicate the procedure of
frequent replanning. Results are as expected quite good
for the machine file approach while higher discrepancies
were found using EPID data. Main problem we face here is
that especially for the machine file based version we do
not have full independent data sources.
EP-1494 The MedAustron proton gantry: nozzle design
recommendations based on Monte Carlo simulations
H. Fuchs
1,2
, L. Grevillot
2
, A. Elia
2
, A. Carlino
2,3
, J.
Osorio
2
, V. Letellier
2
, R. Dreindl
2
, M. Stock
2
, S. Vatnitsky
2
1
Medizinische Universität Wien Medical University of
Vienna, Department of Radiation Oncology & Christian
Doppler Laboratory for Medical Radiation Research for
Radiation Oncology, Vienna, Austria
2
MedAustron lon Therapy Center, Department of Medical
Physics, Wr. Neustadt, Austria
3
University of Palermo, Department of Physics and
Chemistry, Palermo, Italy
Purpose or Objective
MedAustron is equipped with one vertical and three
horizontal fixed beam lines and one proton gantry based
on the PSI gantry 2 design for patient treatments. This
work focuses on simulations and design considerations for
the proton gantry nozzle, allowing an optimization of
beam delivery properties at isocenter.
Material and Methods
Different gantry nozzle designs were evaluated using
Gate/Geant4 Monte Carlo (MC) simulations: air filled
nozzle, helium filled nozzle, full vacuum nozzle, moving
snout, compacting of nozzle elements (vacuum window
and monitors). Design considerations were based on the