Do hardware artefacts influence the performance of head and neck PET scans in patients with oral cavity squamous cell cancer?

OBJECTIVE The purpose of this study was to evaluate the influence of (68)Ge-based and CT-based attenuation correction as well as two standard image reconstruction algorithms on the appearance of artefacts due to dental hardware. Additionally, the intensity of such artefacts was compared with (18)F-fluorodeoxyglucose (FDG) uptake in patients with known oral cavity squamous cell cancer. METHODS Thirty-two metallic and non-metallic objects used for dentistry/dental surgery were scanned in a water-bath filled with FDG on a combined PET/CT scanner. Images were reconstructed with either CT-based or (68)Ge-based transmission data and by using iterative reconstruction or filtered backprojection. The intensity of artefacts was assessed visually using a subjective scale from 0 (no artefact visible) to 4 (very strong artefact), and by quantitative measurements. In a second study, images of 30 patients with known squamous cell cancer and dental hardware were retrospectively analysed by two observers, again using a visual assessment grading system. Wilcoxon signed rank test was used for statistical comparisons. RESULTS Eighteen of 32 objects caused artefacts, which were visible with both attenuation correction methods. CT-based attenuation correction was visually more intense than (68)Ge-based attenuation correction (P<0.0001), and the measured (18)F concentration was also higher (P=0.0002). No difference was found between the reconstruction algorithms. In 28 of 30 patients the primary tumour was visible. FDG uptake in the primary tumour was significantly higher than measured (18)F concentration in artefacts (P<0.0001). CONCLUSION Attenuation correction of PET images generates artefacts adjacent to dental hardware that mimic FDG uptake. In this series, the primary lesion was discriminated from artefacts.