905 / 1082
S889
ESTRO 36 2017
_______________________________________________________________________________________________
to obtain a reliable CBCT based dose calculation. To
validate this approach, we evaluated 11 CT-CBCT
registrations of the head with no visible deformations, and
compared plan calculations on both scans. To assess the
potential to monitor planned dose on the CBCT, 22
patients receiving postoperative head and neck irradiation
with 2 or 3 dose levels were evaluated retrospectively for
a total of 265 CBCT scans. 5 Patients received a new CT
and a replanning during the treatment course. All dose
distributions were evaluated on V95% of the PTV, mean
dose on parotid glands, mandible, oral cavity, larynx,
maximum dose on myelum, and low dose volume (<5Gy).
Results
Validation on 11 patients of the dose calculation showed
an average deviation between planning CT and CBCT scans
of less than 1% on all evaluated dose metrics (Figure 2a).
Evaluation of 22 patients shows deviations of <5% in PTV
coverage in 20 patients over the course of the treatment
(Figure 2b). Two patients showed a higher deviation.
Patient 14 showed anatomical variation that was not
detected during treatment. Patient 18 had a relevant
reduction in PTV coverage during treatment course due to
weight loss and received a new plan. Four other patients
received a replanning because of other considerations,
e.g. a deteriorating condition or treatment side effects.
In the evaluated OAR’s, variations in evaluated metrics of
<5% were observed.
Conclusion
The automated evaluation tool in this study provides a
reliable prediction of delivered dose for the daily patient
anatomy. Evaluation of a series of fractions shows that it
is can detect dose deviations and trigger plan adaptation,
with an action level of approximately 5% deviation in
V95%. Inclusion of deformable image registration is
expected to further increase the reliability of the DVH
predictions.
EP-1660 Improvement in tumour control probability by
adapting dose to daily OAR DVCs
D. Foley
1
, B. McClean
1
, P. McBride
1
1
St Luke's Reseach Oncology Network, Physics, Dublin,
Ireland
Purpose or Objective
A technique using analysis of on-board CBCT images to
adapt the dose to the target on a fraction-by-fraction basis
was developed. This new approach involves using the
upper limit of dose volume constraints (DVCs) as the
objective to be met at each fraction by tracking and
accumulating dose voxels. The aim was to adapt the dose
per fraction such that it was optimised each day without
any organ at risk (OAR) DVCs being exceeded. The impact
on tumour control probability (TCP) and normal tissue
complication probability (NTCP) was evaluated.
Material and Methods
31 patients who underwent prostate treatment were
retrospectively investigated for this study. Initial VMAT
plans consisting of 2 arcs were designed to deliver 74 Gy
in 37 fractions of 2 Gy each to the target. The patients had
on-board CBCT scans taken prior to treatment for between
9 and 33 fractions (436 in total).
An in-house registration algorithm based on phase
correlation[1] was used to retrospectively register CBCT
images to the planning CT to determine the
transformations and deformations in patients’ anatomy.
This allowed the original plan to be recalculated on the
registered CT image that provided the position of the
target and OARs for that fraction. By tracking individual
voxels throughout treatment, the dose was accumulated
and the DVHs and DVC values were determined for each
fraction.