949 / 1082
S933
ESTRO 36 2017
_______________________________________________________________________________________________
was still valid.
Conclusion
The performed optimisation process allowed us to manage
the image quality which met expected quality criteria
with significant reduction in dose.
EP-1724 Phantom image quality evaluation under 3
coil settings for abdominal MR-simulation at 1.5T
O.L. Wong
1
, J. Yuan
1
, S. Yu
1
, K. Cheung
1
1
Hong Kong Sanatoirum & Hospital, Medical Physics and
Research Department, Hong Kong, Hong Kong SAR China
Purpose or Objective
MR-simulation for abdominal radiotherapy often involves
the use of customized immobilization vacuum bags and
radiofrequency (RF) coil holders. Although several types
of RF coils are available for abdominal MR scans, the
influence of different RF coils and settings on image
quality has rarely been studied. In this study, we aimed to
quantitatively compare the quality of image acquired by
three different coil settings for abdominal MR-simulation
scan on a 1.5T MR-simulator.
Material and Methods
A
homogeneous
cylindrical
water
phantom
(diameter~21cm, length~35cm, volume~15L) was
positioned on a flat couch top
with a vacuum-bag. In
combination with a spine coil, three sets of scans, with 4
repeats each, were performed under the coil settings
(Fig1) with either a 18-channel body array (Body18x1), two
6-channel body arrays (Body6x2) or a single 6-channel
body array (Body6x1) on a dedicated 1.5T MR-simulator
(Aera, Siemens Healthineers, Erlangen, Germany). All
images were acquired using a 2D spin-echo T1-weighted
(TR/TE=500/20ms)
and
T2-weighted
(TR/TE1/TE2=2000/20/80ms) sequences (FOV=448mm,
matrix=448x448, slice thickness=5mm, geometric
distortion correction and prescan normalization=ON, 11
slices). For all scans, the coil-to-phantom distance
remained constant by fixing the coil holder height. SNR
was calculated based on AAPM Report 100 using the
central slice from each image set. For image uniformity
assessment, the percent of pixels with intensity within 10%
of the mean signal was calculated as uniformity index (UI).
A rank-sum test was performed to compare SNR and UI
differences between three coil settings.
Results
As illustrated in Fig2, the SNR of Body6x1 (T1:51.2±1.3,
T2:103.8±26.3) was significantly larger than that of
Body18x1 (T1:47.7±1.1, T2:81.9±6.7) for both T1 (P<0.01)
and T2 series (P<0.05). Compared to Body18x1, the SNR of
Body6x2 (T1:46.1±0.9, T2:96.7±10.5) was significantly
lower using T1 series (P<0.05) and larger using T2 series
(P<0.01). Significantly larger SNR of Body6x1 was also
noted comparing to Body6x2 using T1 series (P<0.01). For
image uniformity assessment, UI of Body6x1
(T1:92.8±0.6%, T2:89.0±0.3%) was significantly smaller
than Body18x1 (T1:96.4±0.6%, T2:82.5±0.2%) and Body6x2
(T1:96.0±0.2%, T2:82.6±0.2%) using T1 series (P<0.01),
and significantly larger than Body18x1 and Body6x2 using
T2 series (P<0.01). In terms of SNR and UI, Body6x1
outperformed other two settings for T2-weighted
abdominal MR-simulation. However, shorter coverage
along SI direction and smaller maximum acceleration
factor of Body6x1 might be a limitation for some
applications due to its smaller coil size and fewer array
elements.
Conclusion
Our results suggested that Body6x1 might provide better
SNR and image uniformity for T2-weighted abdominal MR-
simulation scan than other two settings.
EP-1725 Predicting radiation-induced pneumonitis in
NSCLC: a radiobiological and texture analysis study
W. Nailon
1
, W. Lu
2
, D. Montgomery
1
, L. Carruthers
1
, J.
Murchiston
3
, A.W. Yong
3
, G. Ritchie
3
, T. Evans
4
, F.
Little
4
, S.C. Erridge
4
, A. Price
4
, D.B. McLaren
4
, S.
Campbell
4
1
Edinburgh Cancer Centre Western General Hospital,
Department of Oncology Physics, Edinburgh, United
Kingdom
2
School of Engineering University of Edinburgh, Institute
of Digital Communications, Edinburgh, United Kingdom
3
Royal Infirmary of Edinburgh, Department of Radiology,
Edinburgh, United Kingdom
4
Edinburgh Cancer Centre Western General Hospital,
Department of Clinical Oncology, Edinburgh, United
Kingdom
Purpose or Objective
In patients with inoperable non-small cell lung cancer
(NSCLC) reliably estimating susceptibility to radiation-
induced pneumonitis is challenging. Typically dose-volume
histogram (DVH) parameters, normal tissue complication
probability (NTCP) and changes in lung density are used,
however, there is still considerable uncertainty in
predicting individual patient susceptibility. The aim of this
work was to investigate the presence of patient-specific
density patterns that predict the likelihood of pneumonitis
based on image analysis of radiotherapy planning CT
images. The predictive image analysis measures were