Comparison of current reconstruction algorithms for F18-FDG-PET based volume definition – phantom measurements

Ziel/Aim:

Oncological microtherapy planning, e.g. brachytherapy, based on PET-datasets requires exact quantification and volume definition. The aim of this study was to analyze the influence of current reconstruction algorithms on SUVmax, metabolic volumes and spatial resolution using phantom measurements.

Methodik/Methods:

Measurements were performed on a SIEMENS Biograph mCT 64 using a cylindrical phantom (volume, 6595 ml) containing 4 spheres (14-177 ml) filled with F18-FDG (50 kBq/ml) resulting in 3 background fractions (BGF). Images were reconstructed applying 5 algorithms (3D-OSEM, FBP, FBP-TOF, HD-PET, ultraHD-PET) and two slice thicknesses (3 mm, 5 mm). Segmentation of sphere volumes was performed using semiautomatic tumor-to-background algorithm (ROVER). Spatial resolution was determined using point spread function and fitting of radial activity profiles.

Ergebnisse/Results:

The highest SUVmax deviations were found between ultraHD-PET and FBP-TOF with a mean absolute difference of 1.3 (range, 0.7–2.1) and a corresponding mean relative difference of 13.2% (range, 8.2–20.9%). Metabolic volume difference to reference volume was significantly associated to reconstruction algorithm (p<0.05), background fraction (BGF3 vs BGF1, p<0.001; BGF2 vs BGF1, p=0.1), sphere diameter (p<0.001) but not to slice thickness (p=0.3). Best spatial resolution for BGF1 and BGF2 was detected for ultraHD-PET (5.3/4.5 mm) compared to FBP (6.8/6.6 mm) and 3D-OSEM (5.9/5.4). In contrast, similar spatial resolutions were detected for BGF3 ranging from 5.9 mm (3D-OSEM/FBP) to 6.5 mm (HD-PET). However, TrueX-algorithms (HD-PET, ultraHD-PET) showed substantial edge elevations in radial activity profiles (Gibbs artifacts).

Schlussfolgerungen/Conclusions:

Reconstruction algorithms showed substantial effect on SUVmax. This might be of clinical importance if PET quantification is used in oncological microtherapy planning. The clinical impact of high spatial resolution combined with Gibbs artifacts in TrueX reconstructions has to be analyzed in further trials.