Towards Accurate Surgical Navigation Using Intraoperative Imaging and Deformable Models in Trans-oral Surgery

Abstract

Minimally-invasive surgical techniques like trans-oral laser microsurgery and trans-oral robotic surgery have improved patient outcomes in the treatment of oral and throat cancers. Despite these advances, the trans-oral approach still has significant challenges in assessing the extent of the tumor and in locating critical vascular and other anatomic structures beneath the visible mucosal surface. Furthermore, the addition of necessary surgical retractors and laryngoscopes deforms the tumor and critical anatomy from their expected locations, making the surgeons’ task more difficult as they operate with outdated preoperative images.

To overcome these challenges, we have designed and fabricated a 3D-printed polymer laryngoscope and system. Its conception allowed for CT/MR imaging during surgery (intraoperative) with the laryngoscope in place and positioned for larynx exposure. This provided an updated snapshot of how bony and soft tissue move and deform during a typical laryngoscopy procedure, and we have used it to quantify these deformations in patients with and without prior radiation. Intraoperative imaging also enabled us to assess surgeon targeting accuracy in cadaver models and, for the first time ever, provide intraoperative image guidance during trans-oral surgery. We also evaluate predictive deformation models and explore how they might be used to realize surgical navigation without the need for expensive intraoperative imaging. Spatial-tracking and force-sensing capabilities have been integrated into a smart laryngoscope for use during laryngoscopy to provide real-time feedback on the surgical state. These capabilities enable better guidance for the surgeon and serve as input and validation for the deformation models.