Predictive value of high-resolution computed tomography radiomics for assessing the invasiveness of pulmonary adenocarcinoma presenting as ground-glass nodules: A retrospective study.

This study aims to explore the predictive value of radiomics based on high-resolution computed tomography for assessing the invasiveness of pulmonary adenocarcinoma manifested as ground-glass nodules (GGNs). We collected data from 360 GGNs cases confirmed as pulmonary adenocarcinoma by surgical pathology at the First Affiliated Hospital of Chengdu Medical College between January 2019 and December 2023. Clinical information and imaging features were also gathered. The cases were divided into a noninvasive group with 127 cases (35.3%) and an invasive lesion group with 233 cases (64.7%). A total of 279 GGN patients were randomly divided into a training group (195 cases) and a test group (84 cases), with a 7:3 ratio, and 81 cases were included in the validation group. Based on the selected clinical features and radiomics features of GGN, a radiomics model, a clinical model, and a combined clinical-radiomics model were constructed using LASSO and multivariate Logistic Regression. The predictive performance of the 3 models was evaluated and compared using the area under the curve, decision curve analysis, and DeLong test. In the validation set, the area under the curves for the radiomics model (using 29 radiomics features), the clinical model (mean length and air bronchogram), and the combined clinical-radiomics model were 0.856, 0.901, and 0.911, respectively. Decision curve analysis and DeLong's test indicated that the combined clinical-radiomics model has higher value in predicting the invasiveness of GGNs than the other 2 models. The combined clinical-radiomics model demonstrates good performance in predicting the invasiveness of pulmonary adenocarcinoma presenting as GGNs. It has the potential to provide effective imaging methods for patients planning precise treatment.