AI-Assisted 3D Reconstruction for Lung Resection Planning

Patients scheduled for anatomical lung resection who meet the study eligibility criteria will be prospectively enrolled. Preoperative contrast-enhanced chest CT images acquired as part of routine clinical care will be used for 3D reconstruction.

For each patient, an AI-based segmentation system will generate a 3D reconstruction of the relevant pulmonary anatomy, including pulmonary arterial, venous and bronchial structures, as well as lobar and segmental anatomy when applicable. The automated reconstruction will be reviewed without modifying the planned surgical treatment.

AI-generated reconstructions will be compared with expert manual 3D reconstructions and with anatomical findings documented intraoperatively. Particular attention will be given to the identification of surgically relevant bronchovascular branches and anatomical variants.

The primary analysis will assess the proportion of patients in whom the AI-generated reconstruction accurately represents all predefined surgically relevant anatomical structures for the planned resection, without clinically relevant omissions or incorrect structures.

Secondary analyses will assess performance at the individual anatomical-structure level, including arterial, venous and bronchial structures; characterize errors and discrepancies between automated and manual reconstruction; evaluate concordance with intraoperative anatomy; and assess the time required for automated reconstruction and expert review.

The study is a prospective, single-center observational validation study. All diagnostic and therapeutic decisions, including the indication and type of lung resection, will be made according to standard clinical practice independently of study participation.