Modern radiation therapy treatment planning relies on volumetric (CT, or sometimes MRI) imaging to determine tumor location and spread. In addition, radiation oncologists treating head and neck cancer have long used a specific endoscopy - nasopharyngoscopy - to augment the treatment planning volumetric scan. In endoscopy, a thin, flexible light-pipe with a camera is inserted into the patient to obtain optical videos. Endoscopy is done because it is clear to the experienced clinician that direct visualization of a tumor and the surrounding anatomy can reveal valuable information not seen on CT or MRI scans. However, there is currently no way for the clinician to delineate the tumor volume on the endoscopic video and have that information placed correctly into the CT dataset used for treatment planning. Also, there is currently no way to view a combined CT/video dataset. Such a registration has not yet been accomplished in part because of the formidable problems of accurately registering a video taken during nasopharyngoscopy with a planning CT scan. In this proposal we will create a new method to register CT scans with endoscopic videos to produce a fused representation that can then displayed to the clinician to improve radiation treatment planning. To accomplish this, we will further develop and integrate methods of deformable registration, mechanical modeling, motion estimation, computer vision, and 3D interactive display. With such software, radiation treatment planning will have an entirely new and complementary source of information with which to determine tumor extent. Although outside the direct scope of this project, we believe that this technology can be extended to other parts of human anatomy and can be useful for surgical planning as well as radiation treatment planning.
We will develop advanced image processing and display technologies and integrate them in such a way as to allow a clinician to simultaneously view fused internal photographs of the patient's throat and voice box along with the corresponding CT scans. Such a system will result in better tumor targeting for cancer patients undergoing radiation treatment of that part of the body. More accurate treatment will likely result in a highe cure rate for these cancers.
|Zhao, Qingyu; Pizer, Stephen; Niethammer, Marc et al. (2014) Geometric-feature-based spectral graph matching in pharyngeal surface registration. Med Image Comput Comput Assist Interv 17:259-66|