S467
ESTRO 36
_______________________________________________________________________________________________
Image analysis: The two dynamic MR series were analyzed
separately to quantify typical one minute tumor
displacements along two orthogonal directions; superior
(S), inferior (I), anterior (A) and posterior (P). All time-
points affected by non-respiratory associated tongue
motion or deglutition were manually discarded from the
analysis. One time-point was selected as the reference
and all other points were non-rigidly registered to the
reference using a validated optical flow algorithm [1].
Motion fields were computed for all the pixels inside the
tumor and combined into a single distance metric by
assessing the maximum contour coordinates (FIG1-A+B).
Typical 10-minutes displacements were investigated by
computing the difference
Results
Mean maximum 1-minute tumor displacements amounted
to 2.08 (SD 2.34) mm in (S), 2.26 (SD 1.48) mm in (I) and
1.66 (SD 0.93) mm in (A); 1.65 (SD 1.23) mm in (P) (FIG1-
D). However, there was strong inter-subject variability
within the laryngeal and oropharyngeal subgroups, with
laryngeal tumors exhibiting periodic displacements up to
14 mm in (S) (FIG1-C). The typical 10 minute shifts were
smaller than 2 mm for all patients (not shown in figure),
with means values of 0.62 (SD 0.44) mm in (S) ; 0.64 (SD
0.58) mm in (I); 0.49 (SD 0.51) mm in (A); 0.41 (SD 0.36)
mm in (P).
Conclusion
Although tumor displacements were small, there were
three subjects that exhibited resting-state displacements
larger than 5 mm. This suggests individualized ITVs for the
laryngeal tumors and oropharyngeal tumors, instead of
applying 5 mm margins in both I and S directions for
laryngeal tumors. The 10-minutes intrafraction shift was
smaller than 2 mm across all the patients and directions
and did not show any outliers.
PO-0859 Impact of 4DCBCT reconstruction algorithm
and surrogate on motion representation
E. Steiner
1
, C.C. Shieh
1
, V. Caillet
2
, N. Hardcastle
2
, C.
Haddad
2
, T. Eade
2
, J. Booth
2
, P. Keall
1
1
University of Sydney, Radiation Physics Laboratory-
Sydney Medical School, Camperdown, Australia
2
Northern Sydney Cancer Centre- Royal North Shore
Hospital, Radiotherapy Department, St Leonards,
Australia
Purpose or Objective
Lung tumour motion exceeding the observed motion from
planning 4D computed tomography (4DCT) is of concern in
stereotactic ablative body radiation therapy (SABR).
4D cone-beam CT (4DCBCT) facilitates verification of
tumour trajectories before each treatment fraction and an
accurate patient setup. This work aims to assess the
impact of the selection of the reconstruction algorithm
and surrogate for binning on the motion representation in
the 4DCBCTs using implanted Calypso beacons in the lung
as ground truth.
Material and Methods
4DCBCTs were reconstructed from projections for
treatment setup CBCT for 1-2 fractions of 6 patients using
the prior image constrained compressed sensing (PICCS)
method and the FDK (Feldkamp-Davis-Kress) method. For
both methods reconstructions were performed based on
the internal Calypso motion trajectories (three beacons
per patient) or an external respiratory signal (Philips
Bellows belt). The Calypso beacons were segmented for
all 10 bins of the 4DCBCTs and the beacon centroid motion
compared to the motion range from the images. Paired t-
tests were performed on the mean size of excess beacon
motion and the proportion of scanning time with motion
larger than represented in 4DCBCT in order to identify a
superior reconstruction method.
Results
All methods for 4DCBCT reconstruction failed to capture
sudden motion peaks during scanning and underestimated
the actual beacon centroid motion (see Fig. 1), which is a
result of phase-based binning and averaging the images in
a bin. For the SI direction in general, reconstructions using
the belt signal, led to a representation of a larger motion
range (PICCS: 4.88±3.30mm, FDK: 4.81±3.35mm) than the
Calypso-based reconstruction (PICCS: 4.71±3.22mm, FDK:
4.76±3.29mm). However, the difference was not
significant, as for none of the other directions. For
comparison also the Calypso motion during the treatment
exceeding the 4DCBCT motion range is shown in Figure 1.
Figure 1. Proportion of intra-CBCT (solid lines) and intra-
treatment (dashed lines) motion in SI direction larger than
the motion represented in the reconstructed 4DCBCT for
reconstruction with a) PICCS Belt, b) PICCS Calypso, c) FDK
Belt and d) FDK Calypso.
Conclusion
All 4DCBCT reconstruction methods failed to rep resent
the full tumour motion range , but performed similar.
Thus, the belt as an external surrogate is sufficient for
4DCBCT reconstruction. For a safe treatment in spite of
motion exceeding the motion range from the images,
adequate ITV-to-PTV margins or a real-time treatment
adaptation directly tackling motion peaks and
unpredictable motion need to be chosen.
While the 4DCBCT is not able to capture and predict the
whole motion range of a treatment fraction, it serves as a
valuable tool for accurate patient setup.
PO-0860 Characterization of a novel liquid fiducial
marker for organ motion monitoring in prostate SBRT
R. De Roover
1
, W. Crijns
2
, K. Poels
2
, R. Peeters
3
, K.
Haustermans
1,2
, T. Depuydt
1,2
1
KU Leuven - University of Leuven, Department of
Oncology, Leuven, Belgium
2
University Hospitals Leuven, Department of Radiation
Oncology, Leuven, Belgium
3
University Hospitals Leuven, Department of Radiology,
Leuven, Belgium