Abstract
Thoracic surgery encompasses a broad spectrum of procedures with varying levels of risk. Preoperative imaging plays a critical role in evaluating anatomical pathology, but its predictive value for postoperative complications remains underexplored. This study aimed to assess whether specific radiologic features identified before surgery can predict key adverse outcomes, including ICU admission, in-hospital mortality, and length of hospital stay. We conducted a retrospective cohort study of 227 adult patients who underwent thoracic surgeries, including lobectomy, esophagectomy, thymectomy, and mediastinotomy, between 2019 and 2024. Preoperative imaging findings from chest radiographs, CT, PET-CT, MRI, and bronchoscopy were coded and analyzed. Outcomes included ICU admission, in-hospital mortality, and hospitalization duration. Univariate and multivariate logistic regressions were used to assess associations between imaging features and outcomes. Non-parametric tests and visual network plots were also applied. Common imaging findings included emphysema (29.1%), pleural effusion (12.8%), and nodules/metastases (7.9%). ICU admission occurred in 15% of patients, and in-hospital mortality occurred in 7.5%. Certain radiologic features, such as mediastinal lymphadenopathy (OR = 2.03) and nodules/metastases, showed a trend toward increased ICU admission. Conversely, features like bronchogram and no abnormalities were associated with a lower risk. Visual network analyses supported these trends. Preoperative imaging features, particularly those related to mediastinal or tumor burden, may offer predictive value for identifying patients at elevated postoperative risk. Incorporating radiologic markers into preoperative assessment could improve surgical planning and triage for intensive monitoring.