Abstract

Endovascular interventions have been broadly embraced in treating and diagnosing vascular and cardiac diseases. The robotic practice of endovascular procedures has significantly reduced radiation exposure of clinicians and improved the precision, stability and controllability of interventional tool motion. Despite improvements in tools and robotic systems, intraprocedural risks of complications including perforation, dissection, embolization, thrombosis, brain lesions, and stroke are still high. Studies pointed out tool-tissue interaction on the arterial wall as the leading cause of complications and the associated post-treatment risks. Clinicians have limited knowledge of intraluminal interactions taking place through the length of endovascular tools during navigation as well as limited control over such interactions. This research proposes a framework for measurement, monitoring, and control of intraluminal contact force (ICF) of endovascular devices. This thesis presents an image-based sensing solution to estimate multiple contact forces (CF) along the catheter/guidewire and delivers an ICF monitoring system during endovascular navigation. It also proposes a semi-automated robotic framework limiting ICF using image-based control methods.
The proposed sensor-less approach employs a numerical finite element simulation of the tool using image-based data. Real-time image segmentation and tracking algorithms are developed to extract tool shape and compute contact deflections and pose data. An FEM model is built using nonlinear beam elements and image-based pose measurements. The model requires tool flexural rigidity distribution as a fixed-parameter input. Accordingly, a set of experiments are performed to measure the equivalent flexural rigidity along the tool using sequential three-point bending tests. To validate the accuracy of contact force estimations, an experimental setup is prepared with separated contact point phantoms allowing direct CF measurements via a F/T sensor. The results show the effectiveness of the proposed approach in accurately estimating multi-point CFs at the side of an interventional tool. In a second study, the proposed force estimation concept is used to implement an image-based intraluminal tool-vessel interaction monitoring system, which has been tested on teleoperated robotic cannulation of aortic arteries in an anthropomorphic phantom. Moreover, this study compares intraluminal CF with the total force exerted on the vascular phantom to highlight the importance of monitoring local tool-tissue interactions. In the experimental setup, a robotic driver system is designed and fabricated based on the methods in conventional manual navigation. The FEM model is updated to consider large axial, bending and shear deformation. Detection and tracking of contacts within the phantom and computation of tool pose are obtained through a real-time imaging algorithm. The proposed method achieved intraluminal monitoring by tracking multi local ICF during procedure and building ICF contour on the phantom arterial wall. Results suggest that high-risk local overloading may happen even when vascular insertion force is low. The image-based method also computes structural stress of the tool in practice. The proposed online tool-tissue monitoring method delivers insight into the intraluminal interactions and is well-suited for clinician visual guidance, robotic control systems, and research tool design.
The final part of the thesis presents a semi-automated robotic insertion framework to control the intraluminal interaction forces under a prescribed safe reference while advancing the navigation procedure. The method uses the proposed image-based force sensing feedback in a velocity-actuated contact force control loop to perform regulated insertion. An automated retraction-reinsertion feature is developed using visual servoing of the tip to relax the excessive forces and extreme deflection caused by friction build-up. The system switches between ICF control and tip visual servoing to advance the navigation. The proposed automated insertion achieved effective control on intraluminal interaction forces during aortic arteries navigation. Experimental study demonstrates superior performance of the automated framework compared to manual teleoperation modes with and without utilizing visual ICF monitoring. Automated ICF control can minimize the risk of complications and enhance the quality of endovascular procedures. Employing ICF as a visual monitoring also improved the task performance compared to traditional teleoperation in both force and motion metrics.