144 / 1082
S131
ESTRO 36 2017
_______________________________________________________________________________________________
and without a strong magnetic field. A magnetic field of
1.5T was used for PRESAGE® and FOX and 1.0T for BANG
TM
(1.5T was not feasible due to size constraints between the
pole pieces for BANG
TM
). PRESAGE® was irradiated with
doses to 5 Gy, FOX to 8 Gy, and BANG
TM
to 10 Gy.
Calibration curves fitting signal read-out and dose were
compared between 0T and 1.5/1.0T for each dosimeter
type.
Results
The optical signal was analyzed for PRESAGE® and FOX,
and the spin-spin relaxation rate R
2
(=1/T
2
) MR signal was
analyzed for BANG
TM
. For all three types of 3D dosimeters,
the calibration curves were linear. For PRESAGE®, the
percent difference between 0T and 1.5T was 1.5%
measured at the spectral peak of 632 nm; for FOX, there
was a 1.6% difference at 440 nm and 0.5% difference at
585 nm (R
2
= 1.00 for all optical calibration curves). The
greatest percent difference for a given point dose was
5.0% at 2 Gy for PRESAGE®, 2.3% at 6 Gy (440 nm) for FOX,
and 5.6% at 2 Gy (585 nm) for FOX. For BANG
TM
, the
percent difference between 0T and 1.0T was 0.7% (R
2
=
1.00). The greatest percent difference for a given point
dose was 0.3% at 10 Gy for BANG
TM
.
Conclusion
The same doses calculated for 0T were delivered for both
0T and 1.5/1.0T irradiations; the expected dose
difference with the strong magnetic field is up to about
0.5%. Considering this potential dose difference and other
uncertainties, the percent differences in response with
and without strong magnetic field were minimal for all
three 3D dosimeter systems, under 1.6% regarding the full
dose response curves and up to 5.6% for a single dose
point. All three dosimeter systems have already been used
for preliminary investigations on the MR-linac, including
electron return effect studies with an air cavity and
assessing changes in the field edges with and without the
strong magnetic field. This study encourages the
continued use of all three types of 3D dosimeters for MR-
IGRT applications without needing to apply a correction
factor for the signal acquired for any of the above 3D
dosimeter systems.
OC-0259 Online quantitative imaging on the MR-Linac
F. Koetsveld
1
, L.C. Ter Beek
2
, P.J. Van Houd t
1
, L.D. Van
Buuren
1
, U.A. Van der Heide
1
1
Netherlands Cancer Institute Antoni van Leeuwenhoek
Hospital, Radiotherapy department, Amsterdam, The
Netherlands
2
Netherlands Cancer Institute Antoni van Leeuwenhoek
Hospital, Radiology department, Amsterdam, The
Netherlands
Purpose or Objective
Quantitative MR imaging provides information on tumor
physiology and treatment response. For patients treated
in a MR-Linac, it is possible to perform daily quantitative
MR imaging. Online quantitative imaging provides valuable
input for radiomics studies, and can potentially be used
for adaptive dose painting.
Due to time constraints in an online setting, it is
challenging to obtain accurate images that can give
detailed information on tumor environment. Therefore we
assessed the accuracy of four quantitative imaging
sequences on the MR-Linac which are potentially relevant
for radiomics and response assessment: T2 mapping,
Dynamic Contrast Enhancement (DCE) including T1
mapping as input for pharmacokinetic modeling, and
Diffusion Weighted imaging (DWI).
Material and Methods
We compared phantom measurements on the MR-Linac
with two diagnostic scanners: a 3T Philips Achieva dStream
and a 1.5T Philips Achieva.
We tested T2 mapping using a slow but accurate Carr-
Purcell-Meiboom-Gill sequence on the Eurospin T05
phantom (Eurospin TO5, Diagnostic Sonar, Livingston,
Scotland). We compared the results of this sequence with
an accelerated sequence optimized to clinically feasible
acquisition times by reducing the amount of spin echoes.
We tested T1 mapping with an Inversion Recovery series,
which is more accurate but too slow to be used in clinical
practice. We compared this with the fast and clinically
applicable Variable Flip Angle (VFA) T1 mapping
technique.
For DCE we further tested stability of the signal during a
7 min sequence on a phantom containing different
contrast agent concentrations.
For DWI, we tested the accuracy of Apparent Diffusion
Coefficient (ADC) measurements of water at 0
o
C with an
Echo Planar Imaging (EPI) sequence and with a Turbo Spin
Echo (TSE) DWI sequence, which is less susceptible to
geometric distortions.
Finally, to demonstrate the image quality and clinical
applicability on the MR-Linac, we made T1 and T2 maps of
the pelvis on a healthy volunteer.
Results
The phantom results for all four sequences are shown in
figure 1. The accelerated T2 mapping is accurate to within
2% standard deviation on all systems. As expected the VFA
sequence shows a bias of 10-15%. This sequence has similar
precision (within 10% standard deviation) on all three
systems. The DCE sequence on the MR-Linac is shown to
be stable over a 7min long dynamic series like the regular
MR scanners, with a coefficient of variation of < 1% for all
contrast agent concentrations. For diffusion, it shows that
the literature value ADC of water of 1.13*10-3 mm
2
/s (Holz
et al, 2000) can be accurately measured using both EPI and
TSE on the MR-Linac.
Figure 2 shows a T1 and a T2 map of the prostate of a
healthy volunteer, using a scan with 2x2x2 mm voxels.