Intraoperative Pulmonary Nodule Localization with Cone-beam Computed Tomography and Deformable Image Registration

Localization of small pulmonary nodules is challenging during video-assisted thoracic surgery. Current localization methods complicate surgical workflow, increase radiation exposure and introduce additional complication risks. This study evaluates the use of cone-beam computed tomography (CBCT) and deformable image registration to help predict the final location of lung nodules in the intraoperative deflated lung. Porcine and phantom models were developed to simulate the CBCT image acquisition process. To improve the lung registration accuracy, two additions to the Demons registration method were tested, including correcting for lung tissue density differences, and estimating whole lung deflation motion with sampled airway tree landmarks (LDE). CBCT image acquisition during VATS was found to be feasible. The target registration error (TRE) for the unmodified Demons method ranged 13.363-33.974 mm. Only LDE helped significantly reduce TRE by an estimated 7.8 Âą 2.3 mm. The developed registration methods lay the groundwork for future CBCT-based intraoperative nodule localization methods.M.H.Sc.2018-11-07 00:00:0