A multidimensional model incorporating radiomics score and liquid biopsy for the prediction of malignant and benign pulmonary nodules.

Background: As the lesions in pulmonary nodules (PNs) are small and the clinical manifestations lack specificity, the etiology of PNs is complex, predisposing them to misdiagnoses missed diagnoses. Thus, the diagnosis and treatment of PNs remains challenging and an important clinical problem. Methods: This study prospectively enrolled 156 patients with computed tomography (CT)-diagnosed PNs who underwent circulating genetically abnormal cell (CAC) testing between January 2020 and December 2021. We collected data on clinical features closely related to the nature of PNs, such as age, smoking history, and type of nodule. All internal regions of interest (ROIs) of PNs in this study were segmented. Radiomic feature extraction was performed on the ROIs, and a radiomics model was constructed using least absolute shrinkage and selection operator (LASSO) regression to obtain a radiomics score (Rad-score). A comprehensive model combining clinical features, Rad-score, and liquid biopsy was constructed using logistic regression analysis. The diagnostic performance of the model was evaluated using receiver operating characteristic (ROC) curves. Results: In this study, 5 radiomics features were screened for model construction. The area under the ROC curve (AUC) of the radiomics model was 0.844 [95% confidence interval (CI): 0.766-0.915] in the training set. The Rad-score, clinical features, and CAC were further combined to construct a multidimensional analysis model. The AUC of the synthesized model was 0.943 (95% CI: 0.881-0.978) in the training set. Conclusions: A multidimensional model is an effective tool for the noninvasive diagnosis of malignant PNs. The validation and combination of multiple diagnostic methods is a productive avenue of research trend for the identification of malignant PNs.