46 / 1082
S33
ESTRO 36 2017
_______________________________________________________________________________________________
Table 1.
Total number of features per category with a
minimum CCC>0.85 before and after wavelet filtering in
all three datasets. Shape features are not calculated after
wavelet filtering.
Conclusion
We found a high agreement between radiomics feature
stability based on 4DCT and test-retest data in lung
cancer, meaning that 4DCT can be used as alternative to
test-retest to eliminate unreliable features. For
oesophageal cancer a subset of 205 stable CT radiomics
features was identified, of which 64 (31%) were not stable
in NSCLC. This underlines the need for tumour-specific
feature selection prior to model building.
OC-0068 Heterogeneous dose escalation in lung: How
robust are high FDG-uptake volumes during
radiotherapy?
A. Haraldsen
1
, C. Lutz
2
, L. Hoffmann
2
, A. Khalil
3
, D.
Møller
2
1
Aarhus University Hospital, Department of Nuclear
Medicine & PET Center, Aarhus C, Denmark
2
Aarhus University Hospital, Department of Medical
Physics, Aarhus C, Denmark
3
Aarhus University Hospital, Department of Oncology,
Aarhus C, Denmark
Purpose or Objective
The Danish NARLAL 2 Trial (NCT02354274) compares 18F-
fluorodeoxyglucose (FDG)-guided dose escalation to the
current standard radiotherapy (RT) treatment (66Gy/33fx
homogeneously). The dose is escalated heterogeneously
delivering mean doses up to 95Gy/33fx to the high FDG
uptake volumes of the primary tumour aiming to improve
local control for patients with locally advanced Non-small
Cell Lung Cancer (NSCLC). In this study, we investigated
the robustness of these high FDG uptake regions within the
Gross-Tumour-Volume (GTV) during the first weeks of RT
and we calculated the influence of these changes on the
dose coverage.
Material and Methods
We evaluated three successive FDG-PET/CT scans
performed on 20 patients. The baseline scan (PET/CT0)
was dated 14 days prior to RT. The first and second scan
during RT were performed on the same scanner at day 7
(PET/CT1) and 14 (PET/CT2). These two scans were rigidly
registered to CT0 on the primary tumour. A sub-volume,
defined by a threshold of 50% of the Standardized Uptake
Value peak (SUVpeak), was delineated within the GTV on
each scan (V0
50,
V1
50
and V2
50
). The overlap of the volumes
was evaluated with the widely used overlap fraction (OF)
and the OF-boost (see definitions in Fig. 1). The OF-boost
specifies the fraction of respectively V1
50
or V2
50
that is
included in the initial escalation volume, V0
50,
and is thus
more clinically relevant for this trial. The escalated dose
plans were then recalculated on CT2 and the dose to V2
50
was compared to the planned dose to V0
50
on CT0 to
estimate the effect of varying FDG uptake on the dose
coverage.
Results
Median values for OF
0-1
and OF
0-2
were 0.79 [0.56;0.91] and
0.85 [0.58;0.95]. OF-boost
0-1
and OF-boost
0-2
yielded
slightly lower median values of 0,77 [0,52-0,91] and 0.82
[0,58-0,92], respectively.
For 70% of the patients, we found an OF-boost>70% for all
scans (Fig. 2 (a)). These patients showed only small or no
dose decrements to V2
50
(maximum 1,3 Gy), indicating
that a 70% OF-boost is sufficient to maintain the initial
escalation dose. In 30% of the patients, the OF-boost
dropped below 70% (Fig. 2 (b)) and the mean dose to the
high FDG uptake volumes decreased with up to 4 Gy. For
these patients, we found that SUV peak declined to ~2
times background activity, rendering the definition of V1
50
and V2
50
questionable.
Conclusion
For the majority of the patients, the high FDG uptake sub-
volume and the escalated dose level were maintained
during RT. For 30% of the patients, the OF-boost decreased
below 70% and a drop in dose coverage to the high FDG
uptake volumes was observed. For these patients the
definition of a 50% of SUV peak volume during RT was
questionable due to low SUV peak values.