Abstract
BACKGROUND: Accurate preoperative assessment of lymph node metastasis (LNM) is crucial for treatment planning and prognostic stratification in patients with lung cancer. This study aimed to develop and validate a predictive model for LNM using radiomic features derived from non-contrast computed tomography (CT) combined with clinical characteristics. METHODS: A total of 403 patients with pathologically confirmed lung cancer were retrospectively enrolled and randomly divided into a training set (n = 282) and an internal test set (n = 121). In addition,30 lung cancer patients from other hospital were collected as an external test set. Clinical variables were collected, and radiomic features were extracted from non-contrast chest CT images using the Radiomics module of 3D Slicer. Feature selection was performed using least absolute shrinkage and selection operator (LASSO) regression. Multiple machine-learning models were constructed based on radiomic features and clinical features. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), and clinical utility was evaluated by decision curve analysis (DCA). Shapley additive explanations (SHAP) were applied to enhance model interpretability. RESULTS: Lymph node metastasis was observed in 35.5% (143/403) of patients. 2 clinical features and 16 radiomic features most strongly associated with LNM were identified. Among the nine constructed models, the combined clinical-radiomic support vector machine (SVM) model demonstrated the best predictive performance, with AUCs of 0.927 in the training set, 0.852 in the internal test set, and 0.812 in the external test set. Decision curve analysis indicated that the combined model provided a favorable net clinical benefit across a wide range of threshold probabilities. CONCLUSION: The proposed clinical-radiomic model based on non-contrast CT achieved good performance in predicting lymph node metastasis in patients with lung cancer and may serve as a noninvasive tool to assist individualized clinical decision-making.