ESTRO 36 Abstract Book

765 / 1082

S749 ESTRO 36 2017
_______________________________________________________________________________________________
RO clinical trials that were open in 2014 was 20, and in
2016, it was 33. Among these studies, the number of
investigator initiated studies that were open in 2014 was
8, compared to 15 in 2016. In 2016, the MOSAIQ screening
assessment has been completed for 36.6% of new patients
across both sites. Completion of GCP certification by all
radiation oncology staff involved with clinical trials has
reached 100%. The quality of data submission has
improved through accurate collection of data at the
required time points.
Conclusion
This new model of care, tailored to the specific needs of
RO, has resulted in increased clinical trial screening and
participation. The RO clinical trials department has
become the chosen model of care across the local health
district.
Figure 1: Percentage of patients on clinical trials
EP-1417 Clinical evaluation of a fully automatic body
delineation algorithm for radiotherapy
T. Fechter1,2, J. Dolz3, U. Nestle2,4, D. Baltas1,2
1 Medical Center - University of Freiburg, Medical Physics
- Department of Radiation Oncology, Freiburg, Germany
2 German Cancer Consortium DKTK, Partner Site Freiburg-
Germany, Freiburg, Germany
3 École de technologie supérieure, Laboratory for
Imagery- Vision and Artificial Intelligence, Montréal,
Canada
4 Medical Center - University of Freiburg, Department of
Radiation Oncology, Freiburg, Germany
Purpose or Objective
The aim of radiotherapy is to deliver the highest possible
dose to the tumour and spare surrounding healthy tissue.
For high efficacy an accurate delineation of the body
outline on planning CT is crucial. On the one hand for dose
calculation, on the other hand to reduce the delivered
dose to the skin. However, depending on the tumour and
treatment type, positioning markers, catheters, breathing
belt, fixation mattress, table and/or blankets are directly
adjacent to the patient’ skin. Algorithms currently
employed in clinical settings cannot often distinguish
those devices from the patient’s body. Consequently,
these devices are included in the body segmentation
which requires tedious manual corrections. In this work, a
fully automatic algorithm for body delineation that can
handle structures adjacent to the patient is clinically
evaluated for various cancer cases.
Material and Methods
The presented approach is based on a series of threshold
and morphology operations, and it was implement ed using
MITK platform. For evaluation purposes , segm entation
was performed on the planning CT of overall 30 patients:
10 lung cancer patients, 10 patients with a prostatic lesion
and 10 rectum carcinoma patients. CT scans were
acquired on different scanners and with different image
resolutions. Body delineations used for real treatment
planning served as reference contours. Similarity between
reference and generated contours was assessed by
computing the volume ratio (VR), Dice's coefficient (DC)
and Hausdorff distance (HD) to evaluate differences with
respect to volume, overlap and shape, respectively.
Results
The mean VR obtained was 0.99 with a standard deviation
(SD) of 0.006. The average amount of false ly classified
pixels was around 0.3 %. Overlap analysis yielded a mean
DC of 0.99 with SD 0.002 which translates in an excellent
agreement. HD calculation resulted in a mean distance of
6.02 mm and a SD of 3.42 mm. For all cases the algorithm
was able to successfully separate the body from adjacent
parts like breathing belt, blankets, mattress etc. (see
figure).
Conclusion
We have presented a fully automatic algorithm for body
delineation on CT that can handle structures adjacent to
the patient. It has been evaluated in a clinical setting,
showing an outstanding performance. Particularly, 30
clinical cases including several body locations were
segmented. Evaluation demonstrated an excellent
agreement with respect to reference contours. For
segmentation no user interaction is required. Results
suggest the suitability of the algorithm for clinical use with
cases of the tested region between thorax and pelvis.
Future work will explore the use of the algorithm for other
body regions.
EP-1418 RandOmized Study Exploring the combination
of radioTherapy with Two types of Acupuncture
treatment
R. Asadpour1, K. Kessel1,2, D. Habermehl1, T. Bruckner3,
S. Sertel4, S. Combs1,2
1Technical Universy Munich TUM, Department of
Radiation Oncology, München, Germany
2Institute of Innovative Radiotherapy iRT, Department of
Radiation Sciences DRS, Neuherberg, Germany
3Institute of Medical Biometry and Informatics IMBI,
Department of Medical Biometry, Heidelberg, Germany
4Bâtiment hospitalier, Department of ORL and Cervical
Surgery, Lausanne, Switzerland
Purpose or Objective
Acupuncture is known to reduce various clinical signs and
symptoms. Often patients treated with radiation therapy
(RT) suffer from side effects such as fatigue,
nausea/vomiting or reduction of quality of life (QoL). Few
randomized data are available to define the role of
acupuncture in the context of radiation oncology as a