Advancing Image-Based Assistive Technologies In Robotic Surgery: Perception, Targeting, and Planning
Abstract: Robotic surgery holds significant potential to enhance surgic
al precision, reduce invasiveness, and ultimately improve patient outcomes. As medical procedures become increasingly sophisticated, the integration of advanced imaging and robotic systems is critical for further enhancing the capabilities of surgeons. Medical imaging is routinely used by surgeons for planning, diagnosis, and therapeutic purposes, providing crucial information that guides surgical interventions. This dissertation aims to bridge the gap between existing robotic technologies and the demands of modern surgery by advancing image-based assistive technologies. Specifically, it focuses on three components: perception, targeting, and planning. The first component focuses on surgical tool tracking using a deep learning-based method to estimate the 6-DOF pose from image without requiring markers or modifications. Furthermore, the second chapter aims to improve precision and assist the surgeon in MRI-guided Laser Interstitial Thermal Therapy (LITT), emphasizing applicator localization, needle trajectory prediction, and early correction to ensure precise targeting with minimal collateral damage. Finally, a simulation-based trajectory planning system for conformal control of directional hypothermia is presented, in which we first developed a user-interactive planning toolkit and further utilized reinforcement learning to autonomously generate optimal ablation trajectories, thus maximizing therapeutic efficacy and control. Collectively, these contributions aim to enhance the current capabilities of robotic-assisted surgery, with the goal of improving safety, precision, efficiency, and providing enhanced support to surgeons during minimally invasive procedures.
Advisor: Professor Gregory Fischer
Committee: Professor Haichong Zhang, Professor Yihao Zheng, and Professor Zhanyue Zhao
Zoom Link: https://wpi.zoom.us/j/99711751924