ESTRO 36 Abstract Book

945 / 1082

S929 ESTRO 36 2017
_______________________________________________________________________________________________
paper tape (Fig.1.a). Table 1 lists scanning and image
reconstruction parameters used. To further assure
unbiased comparison, we have set geometrical image
parameters (highlighted) to be the same. First we scanned
the phantom on the GE scanner and measured the surface
dose. Table 1 reports averaged surface dose values over
four film strips. On each film strip we have taken an
average dose over a dose profile along 10 cm.
Subsequently, we scanned the phantom 3 times on Philips
scanner, each time adjusting mAs setting, until the
surface dose reached the dose measured on the GE
scanner to within 5%. Then, image quality comparison was
performed using CATPHAN-504 modules in terms of spatial
resolution (Fig.1.b), low contrast detectability (Fig.1.c),
image uniformity (Fig.1.d), and contrast to noise ratio
(Fig.1.e).
Results
The lower part of Table 1 summarizes results of the image
quality comparison. In terms of spatial resolution and low
contrast detectability, it appears that the GE scanner is
slightly better for both Head and Pelvis protocols. On the
other hand, while the uniformity of the images obtained
with Head protocol are slightly better for the GE scanner,
the Philips scanner has better characteristics for the Pelvis
protocol. Also, Philips CT images show significantly less
noise for both scanning protocols. Finally, with regards to
CNR the Philips CT images appear in general to be better
than GE except for high Z material (Teflon) for GE Head
protocol.
Conclusion
The GE CT-simulator demonstrated a slightly better image
quality in terms of spatial resolution and low contrast
detectability, which was expected due to its smaller bore
size, and hence lower impact of scattering on the image
quality, while Philips CT produced images with better SNR.
In the case of CNR values we have found that the Philips
scanner provides images of superior image quality than GE
scanner for Pelvis protocol. These findings can be
explained by the fact that GE uses harder beam quality
(HVL=7.4 mm Al) than Philips (HVL=6.9 mm Al) for Pelvis
protocol indicating a dependence of image quality
parameters on energy spectrum.
EP-1718 Application of motion compensation in 4D CT
of oesophageal cancer.
A. Green1, L. Bhatt2, R. Goldstraw3, H. Sheikh2, M. Van
Herk1, A. McWilliam1
1The University of Manchester, Department 58-
Radiotherapy Related Research, Manchester, United
Kingdom
2The Christie Hospital NHS Foundation Trust, Consultant
Clinical Oncology, Manchester, United Kingdom
3The Christie Hospital NHS Foundation Trust,
Radiotherapy Physics, Manchester, United Kingdom
Purpose or Objective
The use of 4D CT has become widespread for treatment
planning of lung cancers, and motion compensation is
known to be useful for the delineation of targets and
organs at risk. To our knowledge, however, motion
compensation has not yet been used for oesophageal
cancer. Here, as in lung cancers, motion due to respiration
induces image artefacts that can make delineation
difficult leading to lower quality treatments. In this work,
we aim to evaluate the potential benefit of motion
compensated CT in oesophageal cancers.
Material and Methods
Using the ADMIRE (Elekta AB, Stockholm, Sweden) auto-
segmentation tool, all 10 phases of the 4D CT scan were
deformable registered to a reference phase and averaged,
excluding 4 phases displaying the greatest motion
velocity.
GTV delineations on motion compensated images of four
patients were then compared to those performed
according to the SCOPE protocol, in which the GTV is
delineated on the inhale, mid-ventilation and exhale
phases before being combined. Delineated volumes were
evaluated as a surrogate for inter-observer uncertainty, as
higher certainty leads to smaller volumes. In addition, the
volume delineated on the motion compensated image is
compared with the average volume of the SCOPE
structures.
Results
In all cases the volume of the GTV delineated on the
motion compensated image was smaller than the average
SCOPE volume (figure 1). For two of the four patients, the
reduction in volume is significant, however for PT1, the
volume delineated on the motion compensated image was
slightly greater than the mean volume delineated in the
three phases. For the final patient (PT4), the difference is
marginal mainly because an extra tumour extension was
observed on the higher quality motion compensated image
(figure 2), indicating a potential clinical benefit.