ESTRO 36 Abstract Book

169 / 1096

S156 ESTRO 36
_______________________________________________________________________________________________
0.6 mm (LR), and 1.0 mm (CC) (Figure 2).
Conclusion
Our study showed that the motion of lung tumours could
be substantially reduced, but not eliminated, using
visually guided DIBH radiotherapy. Intra- and inter-breath-
hold position uncertainty of the tumour and lymph nodes
were mostly less than 2 mm for visually guided DIBH
radiotherapy of non-small cell lung cancer.
OC-0302 Dosimetric evaluation of a global motion
model for MRI-guided radiotherapy
C. Paganelli1, S. Albertini1, F. Iudicello 1, B. Whelan2, J.
Kipritidis 2, D. Lee2, P. Greer3, G. Baroni1, P. Keall 2, M.
Riboldi 1
1 Politecnico di Milano, Dipartimento di Elettronica-
Informazione e Bioingegneria, Milano, Italy
2 University of Sydney, Radiation Physics Laboratory-
Sydney Medical School, Sydney, Australia
3 Calvary Mater Newcastle, Department of Radiation
Oncology, Newcastle, Australia
Purpose or Objective
MRI-Linac therapy will enable real time adaption of
radiotherapy and is being actively developed by several
academic and commercial groups. To acquire images of
high spatial and temporal resolution, interleaved 2D
imaging is typically used. However, to enable closed loop
adaptive radiotherapy, accumulated 3D dose is required.
A possible way to bridge the gap between 2D and 3D
images is via patient-specific motion models. To date, no
dosimetric evaluation of a global motion model based on
interleaved MRI images has been reported. In this work,
we present the use of a global motion model to
compensate for geometric changes during treatment and
to evaluate dosimetric variations between the delivered
and planned dose distributions.
Material and Methods
4DCT and interleaved sagittal/coronal cine-MRI from a
diagnostic scanner (1.5T) were acquired for a lung cancer
patient. A global motion model was built on the 4DCT
dataset using principal component analysis, and updated
through the use of surrogates derived from in-room cine-
MRI data (tumor, diaphragm and lung vessel motion). An
ITV-based IMRT treatment plan (60Gy in 30 fractions) was
developed on the 4DCT and applied to the model output
for dose evaluation. Validation of the motion model was
performed on a CT/MRI XCAT phantom (1mm resolution),
in which the ground truth CT output of the in-room
scenario was available at the time sample of the simulated
cine-MRI. Analysis of different surrogates as well as their
sagittal/coronal motion components were performed in
terms of both geometric and dosimetric variations.
Results
Based on the phantom data, the accuracy of the motion
model was 1.2mm/1.6mm on tumor/diaphragm.
Dosimetric analysis confirmed the ability of the model to
approximate the ground truth (mean differences of
0.49Gy, 0.12Gy and 0.38Gy for PTV D98, PTV D2 and spinal
cord, respectively). For the patient, variations between
the in-room inhale phase and the corresponding planning
phase were 3.1mm/4.9mm on the tumor/diaphragm. With
respect to the planning inhale CT, the model output CT
presented differences mainly on the diaphragm position
(Figure A). Dosimetric changes with respect to the planned
dose were 3.14Gy, 1.82Gy and 0.42Gy for tumor PTV D98,
PTV D2 and spinal cord, respectively (Figure B and C). The
delivered dose was higher than planned since less motion
was present in the MR images than the planning CT.
Conclusion
We provided a dosimetric evaluation based on a global
motion model for MRI-guidance. The proposed model built
on 4DCT was updated based on interleaved 2D MRI data
and validated using a digital phantom. Dosimetric
variations on tumor were observed in the patient study,
demonstrating the utility and importance of using motion
models for dose accumulation. Future work will include
improvements in the motion model for MRI-guidance and
its application to a larger number of patients.
OC-0303 Evaluation of lung anatomy vs. lung volume
reproducibility for scanned proton treatments under
ABC.
L.A. Den Otter1, E. Kaza2, R.G.J. Kierkels 1, M.O. Leach2,
D.J. Collins2, J.A. Langendijk1, A.C. Knopf1
1UMCG University Medical Center Groningen,
Department of Radiation Oncology, Groningen, The
Netherlands
2The Institute of Cancer Research and The Royal Marsden
Hospital, CR-UK Cancer Imaging Centre, London, United
Kingdom
Purpose or Objective
Proton therapy is a highly conformal way to treat cancer.
For the treatment of moving targets, scanned proton
therapy delivery is a challenge, as it is sensitive to motion.
The use of breath hold mitigates motion effects. Due to
the treatment delivery over several fractions with delivery
times extending the feasible breath hold duration, high
reproducibility of breath holds is required. Active
Breathing Control (ABC) is used to perform breath holds
with controlled volumes. We investigated whether the
lung anatomy is as reproducible as lung volumes under
ABC, to consider ABC for scanned proton treatments.
Material and Methods
For five representative volunteers (3 male, 2 female, age:
25-58, BMI: 19 – 29) MR imaging was performed during ABC
at two separate fractions. The image voxel size was
0.7x0.7x3.0 mm 3. Each fraction consisted of four
subsequent breath holds, resulting in a total of eight MRIs
per volunteer. The interval between fractions was 1-4