Optimised PET reconstruction of the head and neck area: improved diagnostic accuracy

Reconstruction parameters are an important factor in PET image quality. In the head and neck area, where the level of photon attenuation is relatively low, standard whole-body reconstruction (SWR) parameters may lead to suboptimal results. The purpose of this study was to evaluate the impact of optimised head and neck reconstruction (OHR) parameters on image quality and diagnostic accuracy, using pathology as the gold standard.

SWR parameters consisted of 2 iterations, 8 subsets and a 6-mm Gaussian filter. Predetermined OHR parameters were 4 iterations, 16 subsets and a 5-mm Gaussian filter, generating images with increased spatial and contrast resolution but also with increased noise. SWR- and OHR-based FDG-PET images of 28 patients with malignancies in the head and neck area were evaluated for primary tumour and pathological lymph nodes. Diagnostic accuracy was determined by histopathological verification after lymph node dissection.

Using OHR, sensitivity for detection of a primary tumour increased from 92% to 100%. Eleven additional lymph nodes were visualised in eight patients, resulting in an increased sensitivity for lymph node metastases from 11% to 44%. Specificity decreased from 89% to 74% owing to visualisation of small reactive lymph nodes. In total, using OHR, FDG-PET diagnosis improved in six patients (21%) at the expense of three additional false positives for lymph node metastasis (11%). Primary tumour SUV(max) increased by 42%, indicating enhanced contrast resolution.

Image reconstruction adapted to low photon attenuation in the head and neck area may improve image quality and the diagnostic value of FDG-PET, despite a slightly higher false positive rate attributable to the fact that visualisation of FDG accumulation in benign reactive lymph nodes is also enhanced.