940 / 1082
S924
ESTRO 36 2017
_______________________________________________________________________________________________
•
·The manual and automatic contours were
compared after every 5 pts added to the atlas
at a time up to a total of 30.
Results
Comparing the mean scores between initial manual and
the MBS contours (Table1), the manual approach
performed better for spinal cord and parotids with
averaged scores of 8.4 (manual) vs 7.2 (MBS). Standard
deviation showed comparable intra-organs variability. DSC
scores were: 1.0 mandible, 0.71 spinal cord, 0.73 right
parotid, 0.72 left parotid, 0.80 brainstem. All 30 pts were
then used to build a customized atlas. Contours analysis,
tested on 4 new pts, is shown in Figure1. After training,
the performance of the ABAS increased for all the OARs
automatically contoured. Best outcomes resulted for
mandible, spinal cord and brainstem for which the score
and DSC are respectively: 10, 1.0; 10, 1.0; 10, 0.95.
Parotids showed lower results: 7.5, 0.70 for right parotid;
7.4,
0.71
for
left
parotid.
Conclusion
The default MBS tool showed a difference in structure
delineation that does not exceed the inter-clinician
variability. The customized atlas developed reached
performances comparable with the clinical gold standard
for mandible, spinal cord and brainstem. To increase
outcomes, several atlases trained on specific sub-
populations could be created reducing the intra-patient
variability and making results closer to optimal
segmentation. As a next step, the influence of image
quality on automatic segmentation will be analyzed.
EP-1710 Chemical stability of BioXmark® following
normofractionated and single-fraction proton beam
therapy
E. Troost
1,2,3,4,5
, S. Menkel
1
, W. Enghardt
1,4,5
, J. Hytry
5
, D.
Kunath
1
, S. Makocki
1
, A. Hoffmann
1,4,5
, R. Jølck
6
1
TU Dresden- Med. Faculty Carl Gustav Carus,
Radiotherapy and Radiation Oncology, Dresden, Germany
2
German Cancer Consortium DKTK, Partner site Dresden,
Dresden, Germany
3
National Center for Tumor Diseases, Partner site
Dresden, Dresden, Germany
4
Helmholtz-Zentrum Dresden-Rossendorf, Institute of
Radiooncology, Dresden, Germany
5
OncoRay, National Center for Radiation Research in
Oncology, Dresden, Germany
6
Nanovi Radiotherapy A/S, Development, Kgs. Lyngby,
Denmark
Purpose or Objective
Use of solid fiducial markers in proton radiation therapy
has been approached with care as their presence may
cause significant local dose perturbations. Recently, a
liquid carbohydrate based fiducial marker (BioXmark®)
has been introduced with minimal dose perturbation
(relative stopping power = 1.164) and visibility properties
suitable for use in image-guided proton therapy (IGPT).
The purpose of this work was to investigate the chemical
stability of the marker for use in both normofractionated
and single fraction proton treatment regimes.
Material and Methods
Ten identical custom-made cylindrical
polymethylmethacrylate (PMMA) inserts (V = 0.95 mL,
d
outer
= 10.0 mm, d
inner
= 5.0 mm, l = 48 mm) were
prepared. BioXmark® markers (150±30 mg) were added to
the bottom of the inserts and water (700 µL) was added
on top of the markers. The inserts were sealed with a
rubber stopper.
A QA dosimetry phantom was modified to accommodate
four PMMA inserts simultaneously by inserting these
sideway into the proton irradiation field (10 × 10cm)
(Figure 1). Four markers (Group A) were irradiated during
daily QA for a period of 51 days with 43 fractions ranging
from 1.44-1.86 Gy resulting in an accumulated dose of
67.4 Gy. Four other markers (Group B) were irradiated
with a single dose of 155.4 Gy and two non-irradiated
Control markers were kept on site for the duration of the
experiments.
Possible chemical alterations caused by proton irradiation
were
evaluated
by
high-performance
liquid
chromatography (HPLC), electrospray ionization mass
spectrometry (ESI-MS),thin-layer chromatography (TLC)
and visual inspection of the markers and the aqueous
phase above the markers.
Figure 1. QA dosimetry phantom setup for proton
irradiation of BioXmark® inserts.
Results
No visual alterations between markers from Group A+B and
the Control markers were observed. HPLC and TLC analysis
of the markers and the aqueous phase above the markers
from all three groups did not indicate chemical
degradation of the marker materials (Figure 2). This
observation was further supported by ESI-MS analysis,
which showed identical m/z fragments for all three groups
(Figure 2).