The interest in quantification of positron emission tomography (PET) images and its application in radiation oncology for improvement of tumour delineation and biologically adapted radiotherapy increased remarkably during the last years. Modern PET/CT scanners and advanced reconstruction algorithms enabled the extended use of PET imaging.
However, they also introduced a new level of complexity in image quantification and treatment planning. In this work different aspects concerning the integration of PET imaging into radiotherapy were considered. One topic of this thesis was the quantification of PET images, which were acquired at the Siemens Biograph 64 True Point PET/CT scanner. The purpose of this investigation was to assess the influence of different reconstruction algorithms for PET based volume quantification. Measurements were performed using a modified in-house produced NEMA IEC phantom filled with varying activity (18F solution).
Six different signal-to-background ratios (SBR)(2.1, 3.8, 4.9, 6.7, 8.9, 9.4 and without active background) were acquired. The phantom consisted of a water-filled cavity with built-in plastic spheres (0.27, 0.52, 1.15, 2.57, 5.58 and 11.49 ml). The following algorithms were available on the Siemens Syngo workstation: Iterative OSEM (5mm Gauss filter), iterative TrueX (5mm Gauss and Allpass filter) and filtered backprojection. Reconstruction was performed using all available products of iteration (i) and subsets (ss). The recovery coefficient (RC) and the threshold (TH) defining the real sphere volume were determined and compared to the clinical standard setting at our department (4i21ss). Additionally, PET acquisitions of small solid lesions of lung patients were reconstructed using OSEM and TrueX with both filters (4i21ss) and analysed. Remarkable differences were found in the TH and the RCs for the different reconstruction settings. For spheres larger than 2.5 ml a constant TH level was found for a given SBR and reconstruction algorithm (4i21ss) and a mean TH could be approximated by a function inversely proportional to the SBR. The TH for the TrueX, especially with Allpass filter, was significantly lower (up to 40 %) than for the OSEM for all sphere sizes.
The RC for the maximum activity was independent of the SBR for the OSEM.
The true activity could be yielded when using a SBR independent correction factor C, which depended on sphere size and increased exponentially for smaller volumes. The TrueX showed a different behaviour since the maximal RC was dependent on the SBR and overestimated the true activity. Therefore it is recommended to use the mean activity, which is more stable, independent of the SBR and better to reproduce. Regarding different combinations of i and ss, the TrueX was more sensitive to permutations than the OSEM. For patient scans the results for volume and activity were not in agreement for the TrueX and OSEM. By applying an adapted TH for the TrueX, the difference to the OSEM volume could be reduced to 5% at most. The maximal tracer uptake for the TrueX with Gauss and Allpass filter was by 20% and 55% higher than for the OSEM, respectively, and could be reduced by applying activity correction. In conclusion, the TrueX results in good image quality. However, for quantification one must be aware of its strong dependence on reconstruction settings. The second aspect of this thesis was the integration of PET based volume delineation into treatment planning. A treatment planning study was performed for paediatric Hodgkin's lymphoma (PHL) patients. It was the aim of the study to explore the potential of PET based target definition and advanced radiotherapy techniques with respect to therapy-related long-term side-effects. Treatment plans for 10 patients were created. PTV1 was defined according to initial CT based tumour extension. An experimental PTV2 was based on anatomy-related lymph node levels encompassing PET positive nodes after chemotherapy. Investigated treatment techniques (prescribed dose 19.8 Gy) comprised opposed field (2F) and intensity modulated photons (IMXT), single field (PS) and intensity modulated protons (IMPT). PTVs and planning techniques were compared concerning volume, DVH parameters and organ equivalent doses (OED).
The PET based treatment volumes PTV2 resulted in volumes significantly reduced by 69% from 902555 cm3 to 281228 cm3. Using PTV2 instead of PTV1, D2% of the heart was reduced from 14Gy to 9Gy and Dmean of the thyroid decreased from 16.6Gy to 2.7Gy. Low (20 %)-, median (50 %)- and high-dose-volumes (80 %) were reduced by 60% for heart and bones when using PTV2.
Proton techniques reduced the high-dose-volume of the lungs and heart by up to 60 %. IMXT increased low-dose-volumes and OED. PTV2 reduced OED by 5410% for all organs at risk.
In summary, the experimental PTV2 concept has a high impact on the treated volume and consequently on OAR sparing. The combination of such new target volume approach and proton therapy could contribute essentially to new PHL treatment concepts in future clinical studies. In conclusion, the results presented in this thesis demonstrate that the impact of quantitative PET information for radiotherapy is feasible but not straightforward. The target volume difference demonstrated in the Hodgkin treatment planning study has a big impact on the reduction of the dose to the adjacent normal tissue which is particularly emphasized when using highly conformal treatment techniques, such as protons.
However, the properties and reconstruction configuration of the used algorithm should be well known.