Clinical and Radiomic Factors for Predicting Invasiveness in Pulmonary Ground-Glass Opacity.

Abstract

Purpose: To identify the invasiveness of pulmonary ground-glass opacity (GGO) by analysing clinical and radiomic features. Materials and methods: Patients with pulmonary GGOs between 2014 and 2019 were included. Clinical features were collected, and radiomic features were extracted from CT data by 3D Slicer software. Predictors of invasiveness of GGO were selected by least absolute shrinkage and selection operator (LASSO) logistic regression analysis, and receiver operating characteristic (ROC) curves were drawn for each prediction model. Results: A total of 195 GGOs were included in this study. Maximum_diameter, CTR and 2 radiomic features were significant predictors of invasive GGOs. The area under the ROC curve (AUC) for prediction models of clinical and radiomic features were 0.84 and 0.809, respectively, and the AUC for the combined prediction model was 0.848. Finally, a nomogram was established for individualized invasiveness prediction. Conclusions: The combination of radiomic features with clinical features can enable differentiation between preinvasive and invasive GGOs. Maximum_diameter>1.3 cm, CTR>0.27 and a higher waveletLHLglcmIdmn were significant factors for predicting invasiveness.