Purpose: To retrospectively compare the sensitivity of a dedicated fluorine 18 fluorodeoxyglucose (FDG) positron emission tomography (PET)/computed tomography (CT) protocol versus a standard whole-body PET/CT protocol for detection of head and neck cancer, with biopsy and follow-up as reference standards.
Materials and methods: Institutional review board approval and informed consent were obtained for this HIPAA-compliant study. Dedicated and standard PET/CT protocols were performed in a phantom and in 55 patients suspected of having head and neck cancer (28 men, 27 women; age range, 21-79 years). The neck phantom contained four 4.4-9.8-mm-diameter spheres. Standard protocol consisted of a midcranium to proximal thigh emission scan of 2-4 minutes per bed position. Dedicated protocol was an 8-minute head and neck scan. Reconstructed field of view and pixel size, respectively, were 30 cm and 2.34 mm for the dedicated and 50 cm and 3.91 mm for the standard protocol. FDG uptake was evaluated visually and semiquantitatively by using standardized uptake values (SUVs). Mean SUV was compared between dedicated and standard protocols with a t test modified for clustered sampling. Receiver operating characteristic (ROC) curves were calculated. A two-tailed P value was used.
Results: In the phantom study, a larger percentage difference (20%-27%) in sphere-to-background ratios with the dedicated than with the standard protocol was observed for 6.0-9.8-mm spheres. In the clinical study, a total of 149 lymph nodes were identified. Five malignant and six benign lymph nodes (mean diameter, 7.1 mm) were visually identified with the dedicated protocol only. SUVs with the dedicated protocol were significantly higher than those with the standard protocol (P<.001). Area under the ROC curve was 0.94 for the dedicated and 0.92 for the standard protocol (P=.56).
Conclusion: FDG PET with either the standard or dedicated protocol was more sensitive than CT for evaluating head and neck lymph nodes. The dedicated protocol improved the detectability of smaller nodes.
(c) RSNA, 2007.