Texture analysis of dual-phase contrast-enhanced CT in the diagnosis of cervical lymph node metastasis in patients with papillary thyroid cancer

Background Computed tomography texture analysis (CTTA) provides objective and quantitative information regarding tumor heterogeneity beyond visual inspection. However, no study has yet used CTTA to differentiate metastatic from non-metastatic cervical lymph node in patients with papillary thyroid cancer (PTC). Purpose To evaluate the value of texture analysis of dual-phase contrast-enhanced CT images in diagnosing cervical lymph node metastasis in patients with PTC. Material and Methods Metastatic (n = 27) and non-metastatic (n = 32) cervical lymph nodes were analyzed retrospectively. Texture analyses were performed on both arterial (A) and venous (V) phase CT images. Texture parameters, including mean gray-level intensity, skewness, kurtosis, entropy, and uniformity, were obtained and compared between groups. Receiver operating characteristic (ROC) curves analyses and multivariate logistic regression analysis were used in our study. Results Metastatic lymph nodes showed significantly higher A-mean gray-level intensity, A-entropy, and lower A-kurtosis and V-kurtosis (allP < 0.001) than non-metastatic mimics. The ROC curve analyses indicated that A-kurtosis demonstrated an optimal diagnostic area under the curve (AUC; 0.884) and specificity (92.59%), while the A-mean gray-level intensity showed optimal diagnostic sensitivity (90.62%). Multivariate logistic regression analysis showed that A-mean gray-level intensity (P = 0.006, odds ratio [OR] = 24.297) and V-kurtosis (P = 0.014, OR = 19.651) were the independent predictor for metastatic cervical lymph node. Conclusion Dual-phase contrast-enhanced CCTA-especially A-mean gray-level intensity and V-kurtosis-may have the potential to diagnose metastatic cervical lymph node in patients with PTC.