3D computed tomography airway geometry for predicting bronchoscopic accessibility in peripheral pulmonary nodules: a prospective study.

Background: Bronchoscopic diagnosis of peripheral pulmonary nodules (PPNs) using radial probe endobronchial ultrasound (rEBUS) has been widely studied. However, the prospective utility of computed tomography (CT)-derived airway geometrical analysis for predicting rEBUS accessibility remains underexplored. This study aimed to determine whether quantitative CT airway geometric characteristics can improve preprocedural planning by predicting the bronchoscopic accessibility of PPNs. Methods: The accessibility of 219 PPNs in 199 patients to rEBUS was prospectively evaluated as easily accessible or difficult/inaccessible. Preprocedural airway geometry was quantified from individual CT scans and used both to guide the procedure and develop predictive models using logistic-least absolute shrinkage and selection operator (LASSO) analyses. Model performance was assessed using the area under the curve (AUC). Results: Of the 219 PPNs, 182 (83.1%) were easily accessible, and 37 (16.9%) were difficult or inaccessible. The mean age was 68.7 (±10.3) years, the average size of the PPNs was 24.5 (±16.1) mm, the mean distance from the pleural surface was 10.8 (±12.5) mm, and 83.1% showed a 'within' bronchus sign. Airway geometrical features-including more acute bifurcation angles, more sharply curved branches, and more elliptical and narrower lumen shapes-were significant risk factors for limited bronchoscopic access. A composite model incorporating these features with clinical variables achieved the highest performance (AUC =0.84), better than both the clinical model (AUC =0.68) and the airway geometrical model (AUC =0.79). Conclusions: CT-based quantitative airway analysis enhances the prediction of PPN accessibility to rEBUS and could support more accurate procedural planning in clinical practice.