843 / 1096
S827
ESTRO 36
_______________________________________________________________________________________________
poorer to published ones for the rest of OAR, this may be
due to the fact that the majority of our cases with 36 Gy
were children which are precisely the more complex cases
to
optimize.
Dosimetric parameters dependent on D
prescr
are presented
in
Table 1.
Conclusion
A RapidArc planning process for craniospinal axis
irradiation has been implemented with significant
improvement on conformity, homogeneity, feasibility and
efficiency.
EP-1539 Parameters for estimating and controlling
small gaps in VMAT treatments
J. Saez Beltran
1
, V. Hernandez Masgrau
2
1
Hospital Clinic i Provincial, Radiation Oncology
Department, Barcelona, Spain
2
Hospital Sant Joan de Reus, Medical Physics
Department, Reus, Spain
Purpose or Objective
It is well established that dose modelling and calculation
of small fields and small MLC gaps is challenging. Indeed,
VMAT plans with a large fraction of small MLC gaps are
prone to present dosimetric discrepancies due to
limitation of the TPS and uncertainties in treatment
delivery. On the other hand, there is a growing interest in
developing tools to characterize robust class solutions for
plans. Our goal was to study leaf gap distributions in terms
of descriptive variables and complexity indices.
Material and Methods
A total of 276 RapidArc plans optimized with Eclipse v13
were exported in DICOM format for this study comprising
all locations (HN, prostate, gyne, brain, other). The
cumulative histogram of leaf gaps for all plans is shown in
figure 1. A set of MATLAB routines was developed to
perform the analysis. The following parameters were
computed: Total Modulation Index (MIt) (Park et, al),
Modulation Complexity Score (MCS) (McNiven et. al), Beam
Irregularity (BI) (Du et. al), mean gap, median gap and
MU/Gy. All these parameters were evaluated as predictors
of the fraction of small gaps, in particular fraction of gaps
smaller than 20, 10 and 5 mm.
Results
There was a very large variability in the distribution of MLC
gaps involved in VMAT plans. A strong exponential
relationship between the median gap and the fraction of
gaps smaller than 5, 10 and 20 mm was observed (r
2
=0.8,
0.92 and 0.95) (see figure 2). A similar but weaker
relationship was observed for the mean gap (r
2
=0.74, 0.87
and 0.94), the MCS (r
2
=0.55, 0.58 and 0.54) and MU/Gy
(r
2
=0.21, 0.26 and 0.25). Neither the MIt nor the BI
presented a relationship with any of the gap fractions
studied.
For median gaps > 30mm, the fraction of gaps lower than
5, 10 and 20 mm was estimated as 0.13 ± 0.07, 0.20 ± 0.06
and 0.36 ± 0.07, respectively. On the contrary, for median
gaps ~10mm, the fraction of gaps lower than 5, 10 and 20
mm was estimated as high as 0.33 ± 0.07, 0.50 ± 0.06 and
0.72 ± 0.07, respectively.
Conclusion
In general, plan complexity indices exhibit a weak
correlation with the fraction of small gaps in VMAT plans.
Similarly, setting limits on the number of MUs, or MU/Gy
has no clear impact on the fraction of small gaps
generated during the optimization process. A good
prediction of the fraction of small gaps can be obtained
from the median gap of the plan. Thus, tolerance levels
for the fraction of small gaps can be defined in terms of
the median gap of the plan, which can be useful to
generate more robust VMAT plans.