Virtual thoracoscopic imaging improves pulmonary nodule localization: A pilot prospective study highlighting upper- and lower-lobe differences

Abstract Background: Accurate localization of small pulmonary nodules remains challenging during minimally invasive thoracic surgery. This study prospectively evaluated the clinical performance of virtual thoracoscopic imaging (VTI), generated from a computed tomography (CT)-based lung deflation prediction model, for intraoperative localization. Given the differences in lung deformation and visual orientation among lobes in thoracoscopic surgery, the study also examined whether localization accuracy varies between upper- and lower-lobe lesions. Methods: VTI was developed preoperatively by applying a deformation model to routine CT scans. Surgeons used both VTI and conventional three-dimensional (3D)-CT images to predict the pleural surface closest to the tumor and placed an intraoperative marking stitch. The distance between the marking site and the actual tumor center was measured and compared between images, and subgroup analysis was conducted according to the pulmonary lobe. Results: VTI more accurately reproduced intraoperative thoracoscopic anatomy and was associated with smaller localization errors than 3D-CT. For lower-lobe nodules, VTI prediction errors were small (approximately 0.5–1.3 cm), consistent with the relatively minimal visual discrepancy between VTI and conventional 3D-CT under a caudal look-up thoracoscopic view. In contrast, for upper-lobe nodules, particularly those for which pleural surface projection is nonintuitive under a caudal look-up perspective, VTI showed a trend toward improved accuracy, reducing prediction errors to 1.2–1.5 cm, whereas 3D-CT–based predictions often deviated by up to 3.5 cm. Conclusions: Localization accuracy varied by lobes, with the greatest VTI benefit observed for upper-lobe nodules, where pleural surface projection is challenging under a caudal look-up thoracoscopic view using conventional 3D-CT. VTI offers a noninvasive and reproducible adjunct for intraoperative localization, potentially reducing the need for invasive marking procedures and further supporting minimally invasive surgery.