Abstract: Background and purpose: Pre-operative imaging examination is especially critical to the guidance on the selection of surgical methods of papillary thyroid cancer. This study aimed to quantitatively assess the diagnostic value of dual-energy computed tomography (CT) imaging for cervical lymph nodes metastasis in patients with papillary thyroid cancer. Methods: This study was approved by the ethics committee, and all patients were provided with written informed consent. Fifty-five consecutive patients with suspected thyroid cancer were prospectively enrolled, and underwent enhanced dual-energy spectral CT scan from August 2016 to May 2017. The quantitative dual-energy CT imaging and qualitative conventional CT scanning data were analyzed and compared by different groups of radiologists. A region of interest was drawn in the cervical lymph nodes with a maximal short axial diameter of greater than or equal to 5 mm. The normalized iodine concentration (NIC) and the slope of the spectral Hounsfield unit curve (λ_HU) during both arterial and venous phases were measured respectively from iodine overlay images and spectral curves. Two-sample t test was performed to compare quantitative parameters at dual-energy CT between the histopathologically proven benign and metastatic lymph nodes. Receiver operating characteristic (ROC) curves were generated for sensitivity and specificity analyses as well as evaluating the diagnostic value of dual-energy CT imaging and conventional CT features. Results: A total of 212 nodes were detected, including 88 benign and 124 metastatic lymph nodes which were successfully matched and labeled during surgery. By using the conventional CT features including node size, degree and pattern of enhancement, necrosis, as well as extranodal extension for detection of metastatic lymph nodes, sensitivity, specificity and accuracy were 4.0%-79.0%, 63.6%-100.0% and 43.9%-72.6%, respectively. Both arterial and venous phases NIC and λ_HU were significantly higher in metastatic than in benign lymph nodes (all P<0.001). The best quantitative parameter for detection of lymph nodes metastasis was arterial phase NIC with optimal threshold value of 25.8%. The sensitivity, specificity, accuracy, positive predictive value and negative predictive value were 90.3%, 96.6%, 92.9%, 97.4% and 87.6%, respectively. The area under curve (AUC) of arterial phase NIC and λ_HU were significantly higher than those of venous phases (all P<0.001). Conclusion: Compared with qualitative conventional CT features, quantitative parameters associated with NIC and λHU at dual-energy CT showed higher accuracy and better diagnostic performance for cervical lymph nodes metastasis in patients with papillary thyroid cancer.

Key words: Papillary thyroid carcinoma, Lymph nodes, Tomography, X-ray computed