ESTRO 36 Abstract Book

168 / 1096

S155 ESTRO 36
_______________________________________________________________________________________________
provide information on the average tumor position, for
spine and lung SBRT.
Material and Methods
In total, 38 fluoroscopy datasets (1 dataset/arc) of 16
patients treated with spine SBRT were used for full-arc
CBCT reconstruction. The kV images were continuously
acquired at 7, 11, or 15 frames/s with a field size ranging
from 10.5x9cm² to 26.6x20cm² (full field) during
flattening filter free VMAT delivery. For reconstruction, a
standard “spotlight” mode template was modified to suit
our data, i.e. full 360° trajectory, full fan, no filters, and
100 kV. The FDK filtered back projection algorithm was
used to reconstruct the CBCTs and the scans were
matched to the planning CT in Offline Review (Varian
Medical Systems, Palo Alto, CA). For validation purposes,
the resulting match values were compared to the average
spine offset values found using template matching +
triangulation of the individual kV images. For lung SBRT,
limited-arc CBCTs were reconstructed from fluoroscopic
images acquired during irradiation of a lung lesion
embedded in a 3D printed anthropomorphic thorax
phantom and of one patient treated in breath-hold. In
order to determine which arc length is required to obtain
sufficient image quality for reliable CBCT-CT matching,
multiple limited-arc CBCTs were reconstructed using arc
lengths from 180° down to 20° in steps of 20°.
Results
3D spine CBCT-CT registration revealed mean positional
offsets of -0.1±0.8 mm (range: -1.5–2.2) for the lateral, -
0.1±0.4 mm (range: -1.3–0.7) for the longitudinal, and -
0.1±0.5 mm (range: -1.1–1.3 mm) for the vertical
direction. Comparison of these match results to the
average spine offsets found using template matching +
triangulation showed mean differences of 0.1±0.1 mm for
all directions (range: 0.0–0.5 mm). For limited-arc CBCTs
of the lung phantom, the automatic CBCT-CT match
results were ≤1mm in all directions for arc lengths of 60-
180°, but in order to perform 3D visual verification, an arc
length of at least 80° was found to be desirable. 20° CBCT
reconstruction still allowed for positional verification in 2
dimensions. The figure illustrates a limited-arc CBCT over
80° for a phantom and 100° for a patient.
Conclusion
Using standard techniques, we have been able to obtain
CBCT reconstructions of planar kV images acquired during
VMAT irradiation. For treatments consisting of partial
arcs, e.g. lung breath-hold treatments, limited-arc CBCTs
can show the average tumor position during the actual
treatment delivery. It is anticipated that this capability
could be implemented clinically with few modifications to
current treatment platforms. This could substantially
improve positional verification during irradiation.
OC-0301 Target position uncertainty during visually
guided breathhold radiotherapy in locally advanced
NSCLC
J. Scherman Rydhög1, S. Riisgaard Mortensen1, M.
Josipovic 1, R. Irming Jølck2, T. Andresen3, P. Rugaard
Poulsen4, G. Fredberg Persson1, P. Munck af Rosenschöld1
1 Rigshospitalet, Department of Oncology- Section of
Radiotherapy, Copenhagen, Denmark
2DTU Nanotech and Nanovi Radiotherapy A/S,
Department of Micro-and Nanotechnology- Center for
Nanomedicine and Theranostics, Lyngby, Denmark
3DTU Nanotech, Department of Micro-and
Nanotechnology- Center for Nanomedicine and
Theranostics, Lyngby, Denmark
4Aarhus University Hospital, Department of Oncology,
Aarhus, Denmark
Purpose or Objective
The purpose of this study was to estimate the intra- and
inter-breath-hold tumour position uncertainty in voluntary
deep-inspiration breath-hold (DIBH) radiotherapy for
patients with locally advanced non-small cell lung cancer.
Material and Methods
Patients had liquid fiducial markers injected in
mediastinal lymph nodes, and, if possible, in the primary
tumours. Treatment was delivered during DIBH. Anterior
and lateral fluoroscopic movies were acquired in free
breathing (FB) and visually guided DIBH at three fractions
(start, middle and end) during radiotherapy (33 fractions,
2 Gy per fraction) of nine patients with locally advanced
non-small cell lung cancer. Fluoroscopies were acquired
post treatment for two perpendicular gantry angles
(Figure 1). Marker excursions in free breathing and DIBH,
inter-breath-hold position uncertainty, systematic and
random errors during DIBH in each of the three cardinal
directions were investigated using an image based
tracking algorithm, defining the marker template as one
of the images from the middle of the first DIBH
fluoroscopy.
The mean marke r position during each DIBH, relative to a
template frame for the first fluroscopy, was regarded as
each fractions and markers uncertainty during the DIBH. A
systematic error for the patient group was calculated as
the standard deviation (SD) of all these mean marker
positions. The standard deviation of the markers position
within each DIBH was used to quantify the intra-breath-
hold uncertainty (Figure 1). A root mean square (RMS) of
the intra-DIBH SD was calculated to estimate random
errors.
Results
A reduction of 2-6 mm in marker excursion in DIBH
compared to FB was observed in the three cardinal
directions (anterior-posterior (AP), left-right (LR) and
cranio-caudal (CC)). Fourier transformation of the motion
trajectories indicated that the lymph node motion during
DIBH mainly originated from cardiac motion. The
systematic errors during DIBH were 0.5 mm (AP), 0.5 mm
(LR) and 0.8 mm (CC). The random errors during DIBH were
0.3 mm (AP), 0.3 mm (LR), and 0.4 mm (CC). The standard
deviation of the inter-breath-hold shift was 0.8 mm (AP),