ESTRO 36 Abstract Book

912 / 1096

S896 ESTRO 36
_______________________________________________________________________________________________
Purpose or Objective
This pilot study aims to prospectively assess the setup
uncertainty in head and neck (HN) using thermoplastic
mask immobilization on a dedicated 1.5T MR-sim for
radiotherapy based on a cohort of healthy volunteers.
Material and Methods
11 healthy volunteers received a total of 142 scans (each
scanned 4-40 times) on a 1.5T MR-sim in a treatment
position immobilized with customized 5-point
thermoplastic mask with shoulder fixation to simulate HN-
RT fractions. Volunteers were carefully aligned and
positioned by RT therapists using a 3D external laser
system. The imaging protocol consisted of a T1w SPACE
sequence (TR/TE = 420/7.2ms, FOV = 470mm, 256 coronal
slices, isotropic voxel size of 1.05mm, acquisition time
~5min, geometric distortion correction ON). Six Degree-
of-Freedom rigid body registration based on normalized
mutual information (sampling factor = 5%) were used to
pair-wisely register images with respect to the first session
for each subject. Systematic and random errors of
translation and rotation in positional setup were
calculated. One sample t-test and box-plot were used to
evaluate positional variation of translation and rotation
with a significance level of 0.05.
Results
The overall positional setup repeatability results were
presented in Table I. The group mean translation was
<1mm and mean rotation was <1º, respectively. The
systematic error in LR, AP and SI translation was 0.48mm,
0.23mm and 0.53mm, respectively. The systematic error
in roll, pitch and yaw rotation was 0.07°, 0.002° and
0.27°, respectively. The random error in corresponding
direction was 1.83mm, 0.62mm and 1.88mm for
translation, and 0.36°, 0.01° and 1.09° for rotation,
respectively. 20 (~14.0%), 0 (~0%), and 25 (~18.6%) out of
all 142 scan sessions had displacements >1mm in the LR,
AP, and SI directions, respectively. Displacements >2 mm
were seen only in the SI direction in one session (~0.7%).
One-sample t-test showed significance of group mean
error in AP translation (p<0.001) and all rotations (p = 0.02
in roll and p<0.001 in pitch and yaw). The box-plot of
inter-session translation and rotation was shown in Fig.1.
No significant differences were observed in translation in
LR and SI direction, while AP translation was significantly
smaller. This could be partially explained by the fixed AP
level of the MR-sim couch. The yaw rotation was
significantly larger than roll and pitch rotation, which may
result from the relatively inadequate immobilization
performance in this direction, but yet to be further
investigated.
Conclusion
Small head and neck setup uncertainty and thus high
positional repeatability could be achieved on a 1.5T MR-
sim, suggesting the potentials of MRI for precise HN-RT.
EP-1651 Dosimetric impact of rotations correction in
Stereotactic RT. How much a 6DoF couch is useful?
S. Chiesa1, S. Menna2, A.R. Alitto 1, L. Azario2, G.C.
Mattiucci1, M. Ferro 3, M. Massaccesi1, A. Re1, A.
Piermattei2, V. Valentini1, M. Balducci1
1Università Cattolica del Sacro Cuore -Fondazione
Policlinico Universitario A. Gemelli, Radiation Oncology
Department- Gemelli-ART, Rome, Italy
2Università Cattolica del Sacro Cuore - Fondazione
Policlinico Universitario A. Gemelli, Medical Physics
Department, Rome, Italy
3Università Cattolica del Sacro Cuore -Fondazione di
Ricerca e Cura “Giovanni Paolo II”, Radiation Oncology
Department, Campobasso, Italy
Purpose or Objective
Setup accuracy and organ motion control are essential in
stereotactic radiation therapy (SRT) due to the use of
sharp dose gradients and tight margins around the target
volume. PRO-ART Project was designed to evaluate set up
errors and dosimetric impact of rotational patient
positioning correction using a 6-Degree of Freedom (6DoF)
robotic couch.
Material and Methods
Patients with lung, brain and abdominal lesions were
enrolled and immobilized with Uni-frame or trUpoint Arch
for brain, Breast board or Body Pro-Lok for chest and
abdomen (CIVCO support system) lesions. Eclipse™
Treatment Planning Systems (Varian Medical
System®,Palo Alto,CA) was used for dose calculations of
VMAT plans. A daily KV-Cone Beam Computed Tomography
(CBCT) was performed before each treatment fraction and
translational and rotational shifts were identified,
recorded and applied on the Protura TM Robotic couch to
correct the position. Using MIM 5.5.2 software, the
simulation CT was rigidly registered with CBCT,
considered as primary CT, for each fraction. After
registration, translational errors were applied to correct
the CT position, obtaining a tCT, i.e. CT with only
translational errors correction. Then, rotational errors
were corrected too, obtaining roto-translated CT (rtCT).
Reference treatment plan was copied to translated tTP
and to roto-transalted rtTP and dose distributions were re-
calculated, obtaining two treatment plans for each
fraction. DVHs dosimetric parameters were compared.
Results
In this study 179 CBCT were performed on 47 patients (14
with primary tumours and 33 with metastatic lesions) and
358 treatment plans were calculated (179 tTP and 179
rtTP). Geometric and dosimetric analysis are reported in
Table 1. There was no correlation between translational
and rotational shifts. Rotational shifts resulted greater
than 1° in 40% of cases. Average variations in PTV and CTV
V95% were negligible, but we observed variations of PTV
V95% >2% in 30%, 8% and 12% of cases (CBCTs) respectively
for brain, thorax and abdomen patients. OARs proximity
to the target caused variations >2% and rotations around
each axis could determine important changes depending