Abstract: Objective To evaluate the diagnostic efficacy of a combined model based on ultrasound radiomics features and conventional sonographic features in differentiating tuberculous lymph nodes from metastatic lymph nodes, so as to provide more reliable evidence for clinical auxiliary diagnosis. Method A total of 346 patients with tuberculous lymph nodes and 109 patients with metastatic lymph nodes diagnosed in the Fourth People's Hospital of Nanning from January 2020 to December 2024 were retrospectively enrolled. Complete ultrasound images and clinical data were collected. Patients were divided into training set (242 tuberculous lymph nodes, 76 metastatic lymph nodes) and test set (104 tuberculous lymph nodes, 33 metastatic lymph nodes) at a 7:3 ratio using a random number table method. Lesion regions of interest (ROI) were delineated, statistically significant ultrasound radiomics features were screened out to establish the radiomics model. Conventional sonographic features with statistical differences were selected to construct the clinical model. The combined model was established by integrating radiomics features and conventional sonographic features, and its diagnostic performance was assessed. Result Age, gender, minimal short diameter and internal echo showed statistically significant differences between tuberculous and metastatic lymph nodes in both training and test sets(P<0.05). Finally, 8 radiomics features were screened out through LASSO regression. In the training set, the AUC of the combined model was 0.872 (95%CI:0.826-0.918), the sensitivity was 0.730, and the specificity was 0.868. In the test set, the AUC was 0.875 (95%CI:0.805-0.945), the sensitivity was 0.765, and the specificity was 0.833. DeLong test showed that the AUC of the combined model was significantly higher than that of the radiomics model and clinical model in both sets (all P<0.05), indicating favorable clinical application value. Conclusion The combined model constructed by ultrasound radiomics features and conventional sonographic features exhibits satisfactory diagnostic performance in differentiating tuberculous lymph nodes from metastatic lymph nodes.

Key words: Tuberculous cervical lymph nodes, Metastatic lymph nodes, Ultrasound radiomics, Differentiation, Diagnosis