ESTRO 36 Abstract Book

942 / 1082

S926 ESTRO 36 2017
_______________________________________________________________________________________________
Radiation-induced lung fibrosis is an unwanted side effect
of curative radiotherapy. Radiological findings include
changes in the mediastinum such as shift toward the
ipsilateral lung due to treatment related volume loss. In
this work we investigate building a standardized and semi-
automatic method to quantify mediastinum changes as
indicator of lung fibrosis.
Material and Methods
31 patients treated with conventional chemoradiotherapy
were included. This was a sub-group of a non-randomized
phase I/II isotoxic trial which enrolled stage II and III
NSCLC (IDEAL-CRT). Each patient underwent a baseline
PET-CT or diagnostic CT before treatment, and a
diagnostic CT for follow-up after 12 months (average: 433
days, range: 358-496 days). Rotation and thickening of the
anterior junction line were taken as measures of
mediastinal change. β was defined as the angle between
the posterior-anterior direction and the line that connects
the centroid of the spinal canal to the centroid of the
anterior junction line at carina level in the follow-up scan;
similarly, the thickness of the junction (t) was defined as
the minimum distance between the two lungs at the same
level (Fig. 1). Mediastinum shift was then quantified in
terms of the absolute difference between the β measured
on the co-registered baseline and follow-up scans (i.e.,
Δβ= βF- βB), and in terms of the ratio of the thicknesses
(i.e., Δt = tF/tB). This was implemented as a semi-
automatic workflow in Matlab using the open-source
Pulmonary Toolkit
).
Fig.1 – Measures of rotation (β) and thickening (t) of the
anterior junction line.
Results
The changes were characterized using the following
grading systems: for Δβ, 0- no/small rotation (<4°), 1-
moderate (4-8°), 2- large (>8°); and for Δt, 0- no/small
change including moderate shrinkage (<1.2), 1- mild
enlargement (1.2-3), 2- moderate enlargement (>3) (Fig.
2). In the presence of disease or toxicities at the junction,
the definition of its centroid is unclear; hence six patients
were excluded. Significant rotation of the junction line
toward the ipsilateral lung occurred in 64% of the patients
(36% and 28% for grades 1 and 2, respectively). Significant
enlargement of the junction was measured in 56% of the
patients (equal occurrence of grades 1 and 2); shrinkage
was measured in 16% of the cases. There was no
correlation between rotation and thickening of the
junction (ρ=0.17, Pearson’s correlation coefficient).
Fig. 2 – Colour overlay of baseline (red) and follow-up
(blue) scans. Cases of grades (a) 0 and 0, (b) 0 and 2, (c)
2 and 0 for rotation and thickening, respectively.
Conclusion
Chronic lung fibrosis manifests radiologically as
mediastinal shift. We propose a standardized method to
characterize these changes based on the positioning and
thickness of the anterior junction line. Further measures
are needed to fully describe mediastinum change. We aim
to correlate these changes with measures of lung fibrosis.
EP-1713 Feasibility of low dose 4D CBCT in patients
with lung cancer.
A. Bryce-Atkinson1, A. McWilliam1, T. Marchant2, C.
Faivre-Finn1, G. Whitfield3, M. Van Herk1
1University of Manchester, Division of Molecular and
Clinical Cancer Science- Faculty of Biology- Medicine and
Health, Manchester, United Kingdom
2The Christie NHS Foundation Trust, Christie Medical
Physics and Engineering, Manchester, United Kingdom
3The Christie NHS Foundation Trust, Clinical Oncology,
Manchester, United Kingdom
Purpose or Objective
Respiration correlated cone-beam CT (4D CBCT) provides
information on the respiratory motion, providing accurate
tumour localisation in the lung taking into account
breathing motion. This project aims to investigate the
potential for 4D CBCT dose reduction, evaluating its effect
on image quality, amplitude estimation and registration
accuracy.
Material and Methods
4D CBCT images were reconstructed from 9 lung cancer
patients with a minimum of 800 projections (120kV, 16mA,
10-40ms), phase sorted to the nearest projection over 10
phases. The amplitudes of tumour motion ranged from
3.9-20mm. Dose reduction was simulated by
reconstructing using 50%, 25%, 16.67%, 12.5% and 10% of
projections and varying the number of phase bins, sorting
all remaining projections. The image quality of each
reconstruction was assessed visually and by testing
registration accuracy for bone (3D) and tumour (4D).
Registration accuracy was evaluated by comparing results
to the standard reconstruction, for both amplitude of
motion in the superior-inferior direction and correctable
(mean) position.
Results
Reducing the number of projections had little effect on
registration accuracy overall for both bone and soft tissue,
despite very poor visual image quality at low number of
projections. Reducing the number of reconstructed phases
recovered visual image quality (Figure 1) at the cost of
underestimating amplitude and 10 phases yielded best
registration accuracy (Table 1). When using 10% of
projections, motion was reported at less than 5mm for 89%
of patients at 3 and 5 phases. Increasing to 10 phases
recovered the tumour motion to approximately that of the
standard reconstructions. Automatic soft tissue
registration required manual preregistration for two
patients: one consistently across all reconstructions, and
the other only for the 10 phase reconstruction with 10% of
projections.