Investigating Ultrasound-Guided Autonomous Assistance during Robotic Minimally Invasive Surgery

Despite it being over twenty years since the first introduction of robotic surgical systems in common surgical practice, they are still far from widespread across all healthcare systems, surgical disciplines and procedures. At the same time, the systems that are used act as mere tele-manipulators with motion scaling and have yet to make use of the immense potential of their sensory data in providing autonomous assistance during surgery or perform tasks themselves in a semi-autonomous fashion. Equivalently, the potential of using intracorporeal imaging, particularly Ultrasound (US) during surgery for improved tumour localisation remains largely unused. Aside from the cost factors, this also has to do with the necessity of adequate training for scan interpretation and the difficulty of handling an US probe near the surgical sight. Additionally, the potential for automation that is being explored in extracorporeal US using serial manipulators does not yet translate into ultrasound-enabled autonomous assistance in a surgical robotic setting.
Motivated by this research gap, this work explores means to enable autonomous intracorporeal ultrasound in a surgical robotic setting. Based around the the da Vinci Research Kit (dVRK), it first develops a surgical robotics platform that allows for precise evaluation of the robot’s performance using Infrared (IR) tracking technology. Based on this initial work, it then explores the possibility to provide autonomous ultrasound guidance during surgery. Therefore, it develops and assesses means to improve kinematic accuracy despite manipulator backlash as well as enabling adequate probe position with respect to the tissue surface and anatomy. Founded on the acquired anatomical information, this thesis explores the integration of a second robotic arm and its usage for autonomous assistance. Starting with an autonomously acquired tumor scan, the setup is extended and methods devised to enable the autonomous marking of margined tumor boundaries on the tissue surface both in a phantom as well as in an ex-vivo experiment on porcine liver. Moving towards increased autonomy, a novel minimally invasive High Intensity Focused Ultrasound (HIFUS) transducer is integrated into the robotic setup including a sensorised, water-filled membrane for sensing interaction forces with the tissue surface. For this purpose an extensive material characterisation is caried out, exploring different surface material pairings. Finally, the proposed system, including trajectory planning and a hybrid-force position control scheme are evaluated in a benchtop ultrasound phantom trial.