• standardized uptake value (SUV)
• PET/MR
• PET/CT
• Head-and-neck cancer (HNSCC)
• Artifacts

Aim of this study was to evaluate the performance of PET/MR in patients with Head and Neck Cancer including in the evaluation an acquisition protocol which we composed for this new modalities and a comparison of the quantification ability of the new device in terms of standardized uptake value (SUV) and its diagnostic outcome with that of PET/CT. The study population comprised 28 patients with head and neck cancer who underwent a single-injection dual imaging protocol with PET/CT and subsequent PET/MR. PET/CT scans were performed applying standard clinical protocols. Subsequently PET/MR was performed using whole-body Dixon MR-sequence for attenuation correction and a dedicated protocol for the neck. Artifacts and image quality, Intensity of [18F] FDG uptake, delineation, lesion characterization of all primary or recurrent tumours and cervical lymph nodes detected in PET/CT and PET/MR were analyzed and compared. Further SUVs for suspicious lesions in the head and neck and normal cervical muscles calculated in the two different modalities were compared. Histopathology, imaging and clinical follow-up data were used as reference standards for the final lesions' classification.
Artifacts and image quality was better for PET/MR than for PET/CT with statistically significant difference (p=0.004). Mean delineation of primary/recurrent and lymph nodes concerning intensity of FDG uptake was similar for PET/MR and PET/CT, while PET/MR allowed statistically higher delineation of primary/recurrent and metastatic lymph nodes than PET/CT.
PET/MR characterization of primary/recurrent tumour resulted in 100% sensitivity and 75% specificity while sensitivity and specificity of PET/CT was 90.9% and 75% respectively. No substantial differences in sensitivity and specificity between PET/MR and PET/CT were observed concerning characterization of cervical lymph nodes.
Quantitative analysis revealed highly significant correlations between maximum and mean SUVs of all lesions evaluated at PET/CT and PET/MR.
SUVmean and SUVmax were significantly higher in PET/MR than PET/CT for tumour lesions, while were significantly lower in PET/MR than PET/CT for cervical normal muscles.
Conclusion: Simultaneous PET/MR acquisition was feasible and delivered in a reasonable acquisition time high quality, diagnostically sufficient PET and MR data. In particular, our data indicate that the combination of MR and PET was beneficial especially for assessment of the primary or recurrent tumour. Moreover, results of the quantitative analysis show that SUVs derived from [18F] FDG-positive lesions on PET/MR and in normal cervical muscles correlated well with those derived from PET/CT.