Continuum robotic systems for intra-operative MRI-guided interventions

Abstract

Recent advancement in interventional Magnetic Resonance Imaging (MRI) has opened a new realm in image-guided surgery. Although MRI offers possibilities to visualize the confined anatomy through high contrast volumetric scan, the limited bore size and extreme magnetic field physically hinder the surgeon from sufficiently accessing the patient. To this end, much research effort has been expended to the development of robotic systems that can operate under MRI environments. It enables surgeons to perform intervention remotely even when imaging is performed, demonstrating remarkable improvements in surgical precision and safety. However, due to the absence of appropriate tools, its clinical potential only prevails for needle-based interventions such as soft tissue biopsy, which require a straight entry pathway to the surgical site. The overall goal of this thesis is to propose a generic hardware and software framework that integrates intra-op MRI, continuum robot and enhanced human-robot interface to overcome the navigation problem in complex environments. Inspired by biological trunks and snakes, continuum robot designs offer possibilities to traverse confined anatomy and hence enable flexible access routes to surgical sites, e.g. through natural orifices like transoral approach. The design challenge of such a miniaturized, MR Safe, yet dexterous manipulator is addressed by employing advanced Finite Element Analysis (FEA) formulation. To achieve precise robot control in confined surgical site, a novel control framework based on non-parametric localized online learning is proposed. Such technique can learn the inverse model directly without prior knowledge of the robot's structural parameters while adapting to unknown soft tissue interactions. Furthermore, parallel processing of real-time MRI data based on Graphical Processing Unit (GPU) is also investigated to enhance the human-robot control interface, providing instant and clear pathologic indication and detailed structural information. All in all, these proposed techniques have contributed to the development of an integrated robotic catheter platform for MRI environments. Low-friction Hydraulics actuation based on rolling diaphragm sealing is incorporated to provide high performance actuation with negligible influence to MRI quality. To demonstrate its clinical potential, detailed quantitative validations were conducted on groups of subjects. Finally, the thesis concludes by highlighting the future work and potential improvements of proposed techniques.