Abstract
Background/Objective: The aim of the present study is to compare diagnostic accuracies of MRI, PET/CT and fused PET/MRI in the assessment of cervical lymph nodes in patients with head and neck cancer (HNC). Methods: Imaging data of 37 patients who underwent MRI, PET/CT, and surgery at our center were retrospectively merged into PET/MR images. Histopathological results of neck dissections and lymph node resections served as the gold standard. Results: MRI and PET/CT were performed on the same day. The mean interval between imaging and surgery was 20 (±19.5) days. All three imaging modalities identified the same number of true positive and false negative cases, resulting in identical sensitivity estimates of 66.7%. Specificities were 90.9% for MRI, 95.5% for PET/CT, and 100% for PET/MRI. The corresponding positive predictive values (PPVs) were 83.3%, 80.7%, and 81.5%, while the negative predictive values (NPVs) were 80.0%, 90.9%, and 100%, respectively. Ten false results are further analyzed regarding side and level of the affected lymph node, and intersections of the three modalities are displayed. In 12 (32.4%) cases, additional findings are depicted in PET/CT, 5 (13.5%) of which are histologically confirmed to be further malignancies. Conclusions: Software-based PET/MRI is an easy-to-perform procedure and provides valuable clinical information in select clinical questions. Furthermore, whole-body acquisition by PET/CT leads to a notable number of additional malignant diagnoses, which especially favors its use in high-risk patients.