885 / 1082
S869
ESTRO 36 2017
_______________________________________________________________________________________________
EP-1626 Predicting motion of normal tissue using
incomplete real-time data during lung radiotherapy.
L.S.H. Bendall
1
, M. Partridge
1
, M.A. Hawkins
1
, J.
Fenwick
2
1
CRUK MRC Oxford Institute for Radiation Oncology,
Department of Oncology- University of Oxford, Oxford,
United Kingdom
2
University of Liverpool, Institute of Translational
Medicine, Liverpool, United Kingdom
Purpose or Objective
Imaging during radiotherapy treatment has the potential
to increase the accuracy and precision of radiotherapy.
MR-linacs can produce high quality images during
treatment delivery. However, for image acquisition and
analysis to be performed in real-time, fast registration
techniques based on incomplete data is required.
Target motion in lung radiotherapy has been extensively
investigated but motion of surrounding organs at risk
(OARs) remain relatively uncharacterised. Consequences
of irradiating nearby OARs to high doses can be fatal. We
have investigated the bronchial tree with the aim of
characterising the motion of the whole structure
throughout the respiratory cycle on the basis of an imaged
subspace.
Material and Methods
4DCT pre-treatment scans of ten NSCLC patients were
used to assess the motion of the bronchial tree in order to
determine the minimum set of transformations necessary
to describe the motion across the respiratory cycle. The
bifurcation points (BPs) of the main airways were initially
used to characterise the respiratory motion and determine
the minimum number of BPs required to be monitored to
accurately infer the position of all BPs.
The accuracy of using the transformations resulting from
the BP investigation to account for the respiratory motion
of the airway sub-structures between the BPs was assessed
using the Dice similarity index, maximum Hausdorff
distance and the percentage of data points within the
structure that are within 0.2/0.5 cm from the baseline
structure. Further investigation into the optimal
transformation required to minimise the mis-registration
of the airway sub-structures throughout the respiratory
cycle was performed including aligning the centre of mass
of subsections of airways between the BPs.
Results
It was found that when the BPs were regionally paired and
a single transformation applied to both BPs, the mis-
registration errors were reduced from 2.19 to 0.18 cm,
suggesting small differential motion between regionally
paired BPs. This may also indicate that the transform may
reduce the mis-registration of the airway sub-structures
between the BPs.
The BP transformations did not always improve the
registration of the airway sub-structure in terms of
maximum Hausdorff distance. The optimal method found
to minimise the maximum Hausdorff distance was by
aligning the centre of mass of the sub-structures.
Table 1 Range of results from registration techniques.
Conclusion
Affine transformations have successfully been used to
align paired bifurcation points within the bronchus to
within 0.18 cm. This demonstrates that it is possible to
describe the motion of all the BPs on the basis of a spatial
subset of data. Similarly, improvements in the registration
of the airway structure has also been achieved, reducing
the maximum Hausdorff distance from 1.14 cm to 0.56 cm
by using a small subset of information - the centre of mass
of sub-structures - at different phases of the respiratory
cycle.
EP-1627 Anatomical advantages of deep inspiration
breath hold for breast radiotherapy: a geometric
analysis
L. Conroy
1
, E. Watt
1
, S. Quirk
2
, J.L. Conway
2
, I.A.
Olivotto
2
, T. Phan
3
, W.L. Smith
2
1
University of Calgary, Department of Physics &
Astronomy, Calgary- Alberta, Canada
2
University of Calgary, Department of Oncology, Calgary-
Alberta, Canada
3
Tom Baker Cancer Centre, Department of Radiation
Oncology, Calgary- Alberta, Canada
Purpose or Objective
Numerous studies have proven the dosimetric benefits of
deep inspiration breath hold (DIBH) for cardiac sparing in
left breast radiotherapy; however, the anatomical
advantages of this maneuver remain largely
uncharacterized on a population basis. We examine the
motion of the heart with respect to the target breast and
chest wall (CW) between end-exhalation and DIBH from a
tangent
beam’s-eye-view
(BEV) perspective.
Material and Methods
Two computed tomography scans were acquired for 10
consecutive left-sided breast cancer patients: a DIBH scan
and an end-exhalation breath hold (EEBH) scan. Breast and
CW were contoured on the DIBH scan according to RTOG
consensus guidelines. The heart was contoured on the
DIBH scan using a validated heart atlas. Contours were
propagated to the EEBH scan using the MIM Maestro
TM
Adaptive Recontour Workflow and edited where
necessary. For all DIBH and EEBH scans, an in-house
MATLAB program was used to measure the separation
between the heart and CW at an angle approximately
perpendicular to the BEV of the medial tangent beam on
each axial slice. This is the estimated location of the heart
that is at greatest risk of entering the tangent beam.
Breath hold amplitudes were measured as the AP distance
between DIBH and EEBH body contours at the mid-
line/nipple-line intersection.
Results
DIBH causes the CW to move superiorly/anteriorly and the
heart to move inferiorly/posteriorly. This relative motion
facilitates coverage of the whole breast with tangents
while sparing the heart (Figure 1). The median (range)
breath hold amplitudes (difference between EEBH and
DIBH) was 1.3(0.7–2.1) cm in the anterior direction, and
the superior border (base) of the heart shifted 1.0(0.2–2.2)
cm inferiorly. The relative motion of the CW with respect
to the heart in DIBH reduced the amount of heart that
would be at risk of entering the tangent field by 1.5(0.6–
3.8) cm (Figure 1b).The median maximum increase in
heart-CW separation was 1.9(1.2–2.4) cm, at a location
4.8(1.6–6.2) cm inferior to the end-exhale heart base
position (Figure 2, colored circles).
Conclusion
In contrast to dosimetric evaluations, results from
population-based geometric analyses can be applied to a
range of breast treatment strategies (e.g., partial breast
irradiation, modified wide-tangents). This study provides
evidence towards predicting the potential benefit of using
DIBH from a free-breathing scan, and can be used to
inform treatment planning strategies robust to inter- and
intra-fraction motion of DIBH treatments.