Subscribe to RSS
DOI: 10.1055/s-0040-1708276
Effect of Reconstruction Settings on PET-Derived Dose-Volume-Histograms after SIRT of the Liver
Publication History
Publication Date:
08 April 2020 (online)
Ziel/Aim PET/CT allows post-therapeutic determination of the Y-90 distribution, which can be converted to maps of absorbed dose. Several parameters affect the accuracy of activity- and dose-maps. Our aim was to determine the effect of varying PET reconstruction parameters.
Methodik/Methods After SIRT, a PET/CT scan was carried out in 10 patients for determining Y-90 distribution. From reconstructed PET data, dose-maps were derived by dose-voxel-kernel convolution. MR images co-registered to PET/CT data were manually segmented for healthy liver (HL) and tumors (TU), for which dose-volume-histograms (DVH) were calculated.
DVHs were compared for six different PET reconstruction settings. Varied parameters were number of iterations, amount of Gaussian post smoothing, and whether point-spread-function (PSF) was modelled or time-of-flight (ToF) information was applied. Evaluated reconstruction settings were:
A) No PSF, No ToF, 2 iterations (i), 12 subsets (s), 6 mm Gaussian
B) PSF, No ToF, 2i, 12 s, 6 mm
C) PSF, ToF, 1i, 21 s, 0 mm
D) PSF, ToF, 1i, 21 s, 3 mm
E) PSF, ToF, 1i, 21 s, 6 mm
F) PSF, ToF, 2i, 21 s, 6 mm.
Ergebnisse/Results Reconstruction F) led to smallest deviations between injected and PET-derived activity (7.2 ± 5.5 %, ranging 0.5–18.8).
For F), mean absorbed dose to HL was 26.8 ± 9.2 Gy (15.3–40.1) and 81.8 ± 48.2 Gy (29.2–157.5) for TU.
A) and B) led to systematically lower (p < 0.05) mean doses than C-F), for HL, as well as for TU. Mean dose differences between reconstructions were on average 3.9 ± 4.2 % (0.0–15.4) for HL and 6.1 ± 5.5 % (0.0–22.4) for TU.
Corresponding points on DVHs for different reconstructions showed higher deviations than in average case, especially in higher dose regions.
Schlussfolgerungen/Conclusions Accuracy of PET-derived activity could be regarded as sufficient, although some outliers were found. The ToF-technique led to most accurate results.
Differences in mean absorbed dose appeared to be minor. A detailed comparison of DVHs indicated that choice of reconstruction parameters has a strong influence on DVH shape.