New Advances in Imaging-Based Preoperative Prediction of STAS in Lung Adenocarcinoma: From CT and PET/CT to Radiomics...

Lung cancer remains highly prevalent worldwide, with persistently high mortality rates, and postoperative recurrence poses a serious threat to long-term survival. Spread through air spaces (STAS) of tumor cells has been identified as a critical factor influencing recurrence and prognosis in lung cancer. Since its formal definition and classification in 2015, numerous studies have confirmed the significant prognostic impact of STAS. Lung adenocarcinoma, the most common subtype of lung cancer, is particularly challenging due to its high histological heterogeneity and generally poor prognosis, making accurate preoperative assessment of STAS especially crucial. However, the gold standard for diagnosing STAS still relies on postoperative pathology, limiting its clinical utility due to diagnostic delay. Methods combining Computed Tomography or Positron Emission Tomography/Computed Tomography with machine learning have demonstrated outstanding potential in the preoperative prediction of STAS. This systematic review focuses on the current applications and recent advances of imaging plus Artificial Intelligence (AI) in predicting STAS in lung adenocarcinoma, and discusses their role in evaluating personalized treatment models and clinical value. In the future, as AI, multimodal imaging, and big data technologies continue to evolve, noninvasive imaging-based prediction of STAS is expected to become more accurate and widely applicable, promoting personalized and standardized management of lung adenocarcinoma and improving patient prognosis and quality of life.