Abstract:

The resection of lung nodules by video-assisted thoracoscopic surgery (VATS) is an established procedure for the diagnosis and treatment of lung cancer. As the localization of these nodules during surgery is difficult visually and/or by palpation, many adjuvant localization techniques are currently used in clinical practice. Nonetheless, these techniques rely primarily on the placement of physical markers in the nodule before surgery, which has various limitations and may lead to complications. To avoid this, an alternative approach consists in the identification of pulmonary nodules in the preoperative CT image, followed by their intraoperative localization using CBCT imaging. However, during surgery, a pneumothorax (lung deflation) causes significant lung deformation and large tissue density changes that hinder the localization of nodules directly in the intraoperative CBCT. This thesis focused on the compensation of the various lung deformations occurring during VATS as a mechanism for the localization of pulmonary nodules in the intraoperative CBCT image. First, a characterization study allowed to identify the principal factors driving lung deformation during VATS. Then, a hybrid method based on image registration techniques and a poroelastic biomechanical model of the lung was proposed to account for such deformation. A retrospective study comprising 5 clinical VATS cases was performed for the evaluation of the proposed method. The results showed that prediction errors measured at anatomical landmarks were of the order of one centimeter, which suggests the feasibility of the proposed approach as a possible solution for the localization of pulmonary nodules during VATS.