Physica Medica: European Journal of Medical Physics - Volume 32 - Supplement 1 - February 2016

31 / 146

out to be more robust mainly for more complex MLC shape. The results
showed that both tools are able to successfully evaluate leaves position and
collimator rotation angle with accuracy less than 1 mm and 1 degree re-
spectively. In error free pattern, the estimated difference between planned
and measured MLC position was within ±1 mm.
Conclusion: A robust and efficient EPID image analysis tool was devel-
oped to automatically extract an image contour and MLC position. Depending
on the image acquisition device resolution a different edge method should
be applied. The results confirmed that this code is suitable and can be simply
implemented in a QA machine program.
The same procedure could be further developed to check MLC leaves po-
sition during VMAT delivery.
A.86
IMPACT OF AUTO BEAM-OFF AND 4D MODEL AUTOMATIC UPDATE ON
TRACKING ACCURACY OF THE VERO SYSTEM
C. Garibaldi A. Bazani F. Pansin R. Ricotti D. Ciardo S. Com
G. Piperno A.M. Ferrari M. Cremonesi B.A. Jereczek-Fossa
R. Orecchia a Istituto Europeo di Oncologia, Milano, Italy; b Università degli
studi di Milano, Milano, Italy
Purpose: To evaluate the impact of auto beam-off and automatic update
of the 4D model during treatment on the accuracy of dynamic tumor track-
ing (DTT) of the VERO system.
Material and Methods: We evaluated the tracking and prediction errors
(during the 4D model building and treatment (ET, EP) analysing the syn-
chronized log files. We simulated significant variations of the breathing
pattern: baseline drift, change of amplitude and period and introduction
of a phase shift between internal and external movement. We evaluated
ET, considering 3 scenarios: (a) no auto beam-off and no model update;
(b) auto beam-off and no model update; and (c) auto beam-off and model
update. We evaluated the influence of the number of samples after which
the beam is turned off if outside the threshold (2 mm) on ET and irradia-
tion time and how many updates were necessary to re-establish a good
correlation model. The impact of the monitoring frequency (1 or 2 Hz) on
the accuracy of the update of the model was also evaluated. Tests were per-
formed so far with sinusoidal patterns, but patient’s respiratory patterns
will be evaluated to determine the best parameters for the clinical setting.
Results: A total of 171 log files were analysed. The mean values of
EP,4Dmodel, EP,treat, and ET were 0.7 ± 0.5 mm, 0.6 ± 0.4 mm, 0.6± 0.4 mm,
respectively. The auto beam-off does not seem to reduce ET considerably
in case of a large variation of the breathing pattern even reducing the
number of samples from 3 to 1, while the corresponding treatment times
increased significantly (35 vs. 123 s). According to the type of change, up
to 4 updates are necessary to restore a good correlation model before the
treatment can be restarted. Increasing the monitoring frequency from 1 to
2 Hz does not seem to decrease ET when using auto beam off.
Conclusions: The automatic update of the 4D model is a powerful tool to
guarantee the accuracy of DTT without increasing the imaging dose due
to fluoroscopy used to build new 4D models.
A.87
COMPARISON OF FIELD-IN-FIELD TANGENTIAL TREATMENT VERSUS THE
CONVENTIONAL TREATMENT
D. Gaudino *, L. Bellesi, G. Stimato, C. Di Venanzio, A. Mameli, E. Infusino,
E. Ippolito, S. Silipigni, C. Rinaldi, S. Ramella, L. Trodella,
R.M. D’Angelillo. Universita’ Campus Biomedico Di Roma, Roma, Italy
Purpose/Objective: Field-in-field (FIF) technique ameliorates convention-
al planning with tangential fields (TANG) for adjuvant treatments of breast
cancers. It consists of the application of additional fields in order to improve
dosimetric parameters. Either FIF or TANG is evaluated comparing dose dis-
tributions on PTV, OAR and DVH constraints. We evaluated retrospectively
the statistical significance of such differences.
Material and methods: 33 patients were evaluated. Endpoints evaluated
were: V95, V105, maximum dose within PTV, maximum dose, lung
maximum dose, lung mean dose, heart maximum dose, and heart mean
dose. Geometrical misalignment was evaluated by EPI. The baseline FIF was
compared to the TANG plan. FIF was recalculated on TPS incorporating the
misalignment data (FIFErrors). A statistical analysis comparing TANG and
FIFErrors results was addressed by Wilcoxon.
Results: We analyzed misalignment data in 33 patients. Mean values for
FIF and TANG plans were respectively: V95= 98.92 versus 98.25%; maximum
dose = 109.0 versus 110.01%; maximum dose within PTV = 108.32 versus
109.01; V105= 4.01 versus 4.42. The FIF was significantly superior to the
TANG plan for V95 (p = 0.003), maximum dose (p = 0.002), maximum dose
within PTV (p = 0.033); it was not significantly superior for V105 (p = 0.201),
although the mean V105 value was overall inferior for the FIF (4.01% FIF
versus 4.42% TANG).
Themean gainby the adoptionof FIF over the TANGaccounted for V95 = 0.67%;
maximum dose = 1.01%; maximum dose within PTV= 0.69%; V105 = 0.41%.
Once recalculated considering themisalignment it was reduced by 2.98% for
V95, 10.14% for maximum dose; 7.93% for maximum dose within PTV; and
24.39% for V105, respectively. SOFTDISO softwarewas used in order to control
correct positioning of patients day by day. Results were analyzed.
Conclusion: FIF technique optimizes the planning and presents a good ge-
ometrical stability, while the impact on organs at risk requires further
evaluation.
A.88
IMPACT OF THE MLC DELIVERY ERRORS ON PATIENT DOSE FOR IMRT
TREATMENTS: A COMPARISON BETWEEN PLANNED DVH AND
RECONSTRUCTED DVH BASED ON MLC LOG FILE (DYNALOG)
S. Gelosa *, F. Parisoli, P. Lattuada, M. Frigerio, C. Berlusconi,
A. Ostinelli. Azienda Ospedaliera S. Anna, Como, Italy
Introduction: MLC position accuracy during sliding window IMRT treat-
ment was verified by the analysis of DynaLog files to compare planned and
recalculated patient dose distributions by DVHs.
Material and methods: 10 clinical plans (breast, pelvis, prostate, lymph-
nodes) were reconstructed to compare the dose distribution of the reference
plan (clinical plan) with the recalculated one by TPS Eclipse (Varian). Plans
were delivered by Varian Clinac iX and the DynaLog files generated by the
controller were imported in LINACwatch software (QualiFormeD, vers.1.2)
to create a RT plan for each session. These were recalculated on patient
CT with the same dose calculation algorithm (AAA13.0.26) and monitor units
as the reference plan.
Results: PTV DVH percentage variations are less than 1% for D95%, D98%,
D2% and Dm, except for breast treatment involving lymph-nodes where
the maximum variation is 2% for lymph-node PTV. The 90–120% isodose
lines of reconstructed plans are wider, improving target coverage without
substantial variations in OAR DVH.
Reconstructed DVH differences between sessions were negligible, with one
exception where little variation in OAR was observed when the delivery
was interrupted.
Conclusion: In the optimization of breast involving lymph-nodes, the OAR
constraint observance may produce MLC movements hard to reproduce,
explaining the difference in PTV lymph-nodes. If the optimization is not
stressed, DVH variation is nonsignificant. This situation is similar for pelvis
when the objective on bowels is difficult to reach. This software allows to
perform patient specific IMRT QA by comparing DVHs, but it does not replace
pre-treatment dosimetric verifications.
A.89
VERIFICATION OF DOSE DISTRIBUTION FROM CCX RU-106 EYE-PLAQUES
BY USING A MICRODIAMOND DOSIMETER
E. Genovese M. Pimpinella A.S. Guerra V. De Coste M. Marinelli
G. Verona Rinat S. Donatiello C. Orlandi A. Romanzo R. Cozz
V. Cannata a Enterprise Risk Management/Medical Physics, Bambino Gesù
Children’s Hospital, IRCCS, Rome, Italy; b ENEA, National Institute of Ionizing
Radiation Metrology (INMRI) Casaccia, Rome, Italy; c Department of Industrial
Engineering, University of Rome Tor Vergata, Rome, Italy; d Bambino Gesù
Children’s Hospital, IRCCS, Rome, Italy
e26 Abstracts/Physica Medica 32 (2016) e1–e70