Computed Tomography Radiomics for Preoperative Prediction of Spread Through Air Spaces in the Early Stage of Surgically Resected Lung Adenocarcinomas

Suh et al. evaluated CT radiomics for predicting Spread Through Air Spaces (STAS) in early-stage lung adenocarcinoma using preoperative CT scans. They retrospectively analyzed 521 patients (550 lesions) who underwent surgical resection. To build a predictive model, they extracted radiomics features from CT images using **AVIEW Research** software for semi-automated 3D tumor segmentation, ensuring precision by excluding large vessels and bronchioles. A radiomics score (Rad-score) was developed from selected features and combined with conventional clinical variables (e.g., lesion type, solid portion size) to create a comprehensive prediction model. Results showed that the combined model outperformed conventional clinical models, achieving higher AUC values in predicting STAS, particularly in temporal validation datasets. High Rad-scores were also associated with lower recurrence-free survival, indicating a poorer prognosis. This study highlights radiomics' added predictive value for STAS, with AVIEW aiding in accurate segmentation for robust preoperative assessment and potentially more personalized surgical planning.