Research on statistical model based surgical planning and advanced registration methods for image-guided surgery

Abstract

Image-guided surgery (IGS), which modernizes surgery and makes it safer and less invasive, has been receiving a lot of interest in recent years. In the preoperative context, diagnostic imaging, such as computer tomography (CT) and magnetic resonance imaging (MRI) provides a foundation for surgical planning including the delineation of surgery target, the surrounding anatomies, and the surgical strategy or trajectory to the target. Additionally, using a statistical atlas in the preoperative stage provides a comprehensive anatomical reference that helps surgeons to better understand complex anatomical variations in a large dataset and spatial relationships between involving individuals, therefore allowing for automatic personalized surgical planning. Such preoperative images and surgical planning information are registered to intraoperative coordinates using a navigation system such as Optical Tracking System (OTS) to visualize tracked instruments in relation to preoperative images. This dissertation seeks to solve three interesting research questions in the preoperative and intraoperative process. The dissertation includes work encompassing: (1) proposing a framework for multimodal image registration between CT and MRI; (2) proposing a computer- assisted surgical planning method using statistical atlas and applying to the humerus surgery scenario; and (3) proposing a dynamic touchable region model applied to a framework for markerless registration in the intraoperative stage.