To analyze the ultrasound features of cervical lymph nodes without perfusion defects, to develop a predictive model for benign and malignant lymph nodes, and to assess the diagnostic value of the model for benign and malignant lesions.

A retrospective analysis was performed on 57 patients with 61 cervical lymph nodes at the Affiliated Hangzhou Chest Hospital, Zhejiang University School of Medicine during January 2013 to May 2020. All lymph nodes were examined by conventional ultrasound and contrast-enhanced ultrasound (CEUS), which were cervical lymph nodes without perfusion defects. According to the results of pathology and Gene X-pert MTB/RIF examination, the lymph nodes were divided into either a benign group (45 cases) or a malignant group (16 cases), and the long diameter/short diameter (L/S) ratio of lymph nodes, cortex, hilum, calcification, echo, and color Doppler flow imaging (CDFI) and CEUS perfusion characteristics were compared between the two groups. Univariate χ² or Fisher test and multivariate Logistic regression analysis were performed to identify the risk factors for malignant lymph nodes to build a predictive model. ROC curve analysis was performed to assess the performance of the Logistic model.

Univariate analysis showed significant differences in the following parameters between benign group and malignant group: L/S ratio (χ²=5.785, P<0.05), hilum (χ²=4.419, P＜0.05), calcification (P<0.05), CDFI (χ²=16.775, P<0.001), perfusion pattern (χ²=13.090, P＜0.001), and enhancement of the hilum (χ²=15.993, P<0.001). Binary Logistic regression analysis showed that non-portal type of CDFI, centripetal perfusion mode, and non-enhancement of the hilum were risk factors for malignant lesions, and the three characteristic variables were entered into the Logistic regression model to build a predictive model as Y=-4.732+1.795X₄+1.989X₅+2.734X₆. The area under the ROC curve of this model was 0.921 (95% confidence interval: 0.846-0.996). The sensitivity, specificity, positive predictive value, and negative predictive value of the predictive model in the differential diagnosis of benign and malignant diseases were 81.3%, 93.3%, 81.3%, and 93.3%, respectively.

The conventional ultrasound and CEUS feature prediction model has high diagnostic value in differentiating benign from malignant cervical lymph nodes without perfusion defects.