889 / 1082
S873
ESTRO 36 2017
_______________________________________________________________________________________________
interface.
Conclusion
This low-cost, computer-vision system for real-time
motion monitoring of the irradiation of breast cancer
patients showed submillimetric accuracy and acceptable
latency. It allowed the authors to highlight differences in
surface motion that may be correlated to tumor motion.
Madibreast detects and tracks accurately external motion
on the breast using low-cost material and accessible open-
source, high-level computer vision libraries. It allows
immediate monitoring by visually displaying an immediate
trace, which can alert that substantial motion could have
occurred. Limitations include some remaining failures of
the apparatus under low-exposure conditions, as well as
considerable CPU occupation.
EP-1633 Respiratory Motion Analysis using a Surface
Guided Radiation Therapy System for Lung SBRT
Patients
M. Jermoumi
1
, D. Cao
1
, V. Mehta
1
, D. Shepard
1
1
Swedish Cancer Institute, Radiation Oncology, Seattle,
USA
Purpose or Objective
Surface guided radiation therapy (SGRT) uses a
camera/projector pair to create a 3D map of a patient’s
surface. SGRT can be used to assist in patient set up, real
time motion monitoring, and respiratory motion
management. In this work, we used SGRT to track the
respiratory breathing pattern for lung SBRT patients. An
excursion gating approach was employed where the beam
delivery was interrupted if the breathing deviated from
the expected pattern. The purpose of this work is to
evaluate the patients breathing motion during SBRT
treatment.
Material and Methods
To date, 10 NSCLC patients have been enrolled in this
study and treated with stereotactic body radiation therapy
(SBRT) using a 12Gyx4 fractionation. Prior to each
fraction, each patient was aligned using SGRT. Next, a
4DCBCT scan was acquired to align based on internal
anatomy. A virtual respiratory tracking point was then
placed close on the patient’s surface close to the sternum.
The patient’s gating window was set based on the end-to-
end amplitude measured during the acquisition of the CT
at the time of simulation. The gating window was
expanded 5 mm beyond the upper level window and 5 mm
below the lower level window. In-house developed code
was used to evaluate the respiratory data collected from
all 40 fractions. The evaluation included an examination
of the end-to end amplitude, the breathing period, and
baseline drift. The correlation between baseline drift and
the treatment time was assessed over the course of
treatment.
Results
The mean (± SD) treatment time was 5.3 (± 1.34) minutes.
The mean (± SD) end-to-end amplitude observed due to
inter-fraction and intra-fraction motion were 6.79(±2.51)
mm and 6.79(±2.86) mm respectively and the mean (± SD)
breathing period was 4.08(±0.44) s. The coefficient of
variance (CV) of the end-to-end amplitude was less than
10% for 50% of the patients and greater than 20 % for 40%
of the patients. In 80% of the treatments, the CV of the
breathing period was less than 10%. A baseline drift of
greater than 2 mm, 3 mm, and 5 mm was observed for
85%, 4%, and 1% of the total treatment times,
respectively. The variability (1SD) of baseline drift was
within a range of 0.49 to 1.34 mm. The baseline drift
versus time showed no correlation (r=0.009, p=0.24).
Conclusion
SGRT provides an excellent tool to track the respiratory
signal of lung SBRT patients. The amplitude variability is
less than 5 mm which is consistent with other reported
studies. These results can be considered as reference data
for decision making for subsequent SBRT lung patients.
EP-1634 Combined 4D and 3D cone beam CT protocol
for lung SBRT for reliable and fast position verification
W. Woliner-van der Weg
1
, N. Gelens
2
, V.H.J. Leijser-
Kersten
1
, P.M. Braam
1
, J. Bussink
1
, M. Wendling
1
1
UMC St Radboud Nijmegen, Radiation Oncology,
Nijmegen, The Netherlands
2
Fontys Paramedische Hogeschool, Medisch
Beeldvormende en Radiotherapeutische Technieken
MBRT, Eindhoven, The Netherlands
Purpose or Objective
In our standard lung SBRT position verification protocol,
the use of online 3D or 4D cone beam CT (CBCT) is based
on the amplitude of tumor motion as measured on 4D
planning CT. This results in about 60% of the patients
having 4D CBCT position verification. While 3D CBCT takes
only 1 min, 4D CBCT lasts about 4 min. With repetitive
imaging, this difference considerably contributes to the
time needed for position verification, and the time the
patient lies on the treatment couch.
We reconsidered our position verification protocol, to
expand the use of 3D CBCT while maintaining reliable
position verification for all patients. Therefore, we
developed a decision protocol, in which 4D CBCTs of the
first treatment fraction are used for further stratification.
Material and Methods
For both 3D and 4D CBCT, the first CBCT has to be within
1 mm in all 3 directions compared to the planning CT,
otherwise the positioning error is corrected with the
treatment couch and a second CBCT is made for
verification. The verification CBCT has to be within 2 mm
in all 3 directions, otherwise the procedure is repeated.
Initial selection for 3D or 4D position verification in our
department is based on the 3D amplitude of tumor motion
measured on 4D planning CT. Patients with a tumor motion
vector length >5 mm are positioned based on 4D CBCT.
In the new protocol the choice for 4D CBCT is reconsidered
during the treatment course. After the first fraction, the
4D CBCTs are also matched in 3D with the planning CT.
Corrections resulting from this match are compared to the
corrections resulting from the initial 4D match. If the
difference is within 0.5 mm in all 3 directions, for the
second and third fraction only the first CBCT is made in
4D, and verification CBCTs are made in 3D. If during the
second and third fraction the difference between the 4D
and 3D match remain within 0.5 mm in all 3 directions, for
the remainder of fractions only 3D online CBCTs are made