![Show Menu](styles/mobile-menu.png)
![Page Background](./../common/page-substrates/page0960.png)
S944
ESTRO 36 2017
_______________________________________________________________________________________________
feasibility of MRI-based atlas-based autosegmentation for
organs-at-risk (OAR) and prostate target volumes, and to
compare the segmentation accuracy with CT-based
autosegmentation.
Material and Methods
Images were retrospectively selected from 6 prostate
patients who received whole field T2 weighted 3D SPACE
MRI and CT in the radiotherapy treatment position (at the
Northern Centre for Cancer Care, Newcastle). Organs at
risk (Bladder, rectum, seminal vesicles, left and right hips,
penile bulb) and the prostate were manually delineated
on the CT and MRI separately. A ‘leave one out’ approach
was used to automatically segment structures onto the
remaining images separately for CT and MRI. Contour
comparison was performed using the DICE index and mean
distance to conformity (MDC) positional metrics. MDC,
DICE and absolute volume were used to assess the
performance of the contouring by comparing the
automatic to the manual contours. A paired t test was
used to determine the statistical significance between MRI
and CT.
Results
The volume analysis (data not presented) showed that
manual and automatic contouring on MRI gave smaller
contours than CT (significantly so for the hips, prostate
and seminal vesicles). The positional analysis results are
shown in table 1. MRI autocontouring was more accurate
than CT for the bladder (MDC significantly so) and the
prostate/penile bulb (although not significantly). There
was little difference in accuracy between CT/MRI
autocontouring for both hips, rectum and seminal vesicles.
Conclusion
Accurate atlas-based automatic segmentation of
structures for prostate radiotherapy is feasible using T1-
MRI; segmentation of the penile bulb and seminal vesicles
was found to be poor. Comparison with CT-based
automatic segmentation suggests that the process is
equally or more accurate using MRI. Although this study
was on a small sample size these results support further
translation of MRI-based segmentation methodology into
clinical practice.
EP-1740 Nationwide audit of multileaf collimators
performance
K. Chelminski
1
, W. Bulski
1
1
The Maria Sklodowska-Curie Memorial Cancer Center,
Medical Physics Department, Warsaw, Poland
Purpose or Objective
The delivery of accurate intensity-modulated radiation
therapy (IMRT) or stereotactic radiotherapy depends on a
multitude of steps in the treatment delivery process. The
proper intensity modulation depends on the proper
functioning of a multileaf collimators (MLC). The aim of
this audit was the control of the proper collimator leafs
positioning.
Material and Methods
The methodology of the audit of small field output
performance was established within the framework of the
CRP E2.40.16 project "Development of Quality Audits for
Radiotherapy Dosimetry for Complex Treatment
Techniques", run by the Health Section of the
International Atomic Energy Agency (IAEA). The
participants of the audit were obliged to irradiate
provided dosimetric films, in a slab phantom, for a specific
leaf arrangement, producing a pattern of five stripes,
commonly called a picket fence. The participants had to
programme such a pattern so that the stripes are 5 mm
wide and are 3 cm distant between themselves. The
Gafchromic EBT2 radiochromic films were placed in a slab
phantom close to maximum dose depth. The irradiation
was 250 MU per stripe.
Results
Thirty two Polish radiotherapy centres took part in the
audit. They were equipped with various accelerator types
and various treatment planning systems. In all cases the 6
MV quality beams were used. The discrepancies between
measured and expected stripe positions were in the range
1.2 mm. For particular participants, the leaf position
discrepancies were in the range -0,5 mm to 0,5 mm. For
particular participants, the mean opening width measured
with films for each pair of leafs was between 6 and 8 mm.
Conclusion
In the audit, the best performance showed the new type
multileaf collimators with 120-160 leafs, The worst
performance showed collimators MLC80 from Elekta. The
results of the audit are very useful for the participants
who should carefully investigate the performance of their
multileaf collimators.
EP-1741 Commissioning of a robotic patient positioning
system equipped with an integrated tracking system
A. Ableitinger
1
, A. Utz
1
, A. Zechner
1
, S. Vatnitsky
1
, M.
Stock
1
1
EBG MedAustron GmbH, Strahlentherapie, Wiener
Neustadt, Austria
Purpose or Objective
Robotic patient positioning systems (PPS) used in clinic
must consider weight-induced couch bending and must
show high reproducibility and stability to achieve the
required positioning accuracy. Extensive commissioning of
these robotic systems is therefore crucial. The aim of the
current work is to determine the positioning accuracy of
the PPS, that is equipped with an integrated optical
tracking system.
Material and Methods
Three different aspects of the PPS were investigated in
this study: the basic characteristics including couch
bending, reproducibility and stability; the relative spatial
deviation in terms of rotation and couch height and the
absolute accuracy of the treatment couch.
The treatment volume of the PPS has a dimension of
115cm x 50cm x 40cm. The robotic system enables couch
rotations of more than 190°, pitch and roll of ±3° and non-
isocentric treatment positions. A photogrammetric
camera tracks the treatment couch of the PPS via
reflecting markers mounted on the bottom side of the
couch (see Fig.1). An iterative position correction loop
aligns the couch to the prescribed position.
The reference instrument was a laser tracker with
reflecting probes. Drilling holes near the indexing
positions (H4-F9) located laterally along the couch every
14cm served as measurement positions for the evaluation
of the basic characteristics and the absolute couch
position. For the relative deviation the drift of one
measurement point on the couch was evaluated.
To determine the bending of the couch it was loaded with
six different weights up to 156kg. The reproducible
positioning for the same couch position (different axis
setting) and the couch stability after 1 hour were
measured with the highest payload.
The evaluation of the absolute spatial deviation was based
on six measurement points being closest to the room
isocenter. They were compared with their expected
coordinates for 1020 different robot positions, poses and
payloads.