Towards Autonomous Navigation of Neuroendovascular Tools for Timely Stroke Treatment via Contact-aware Path Planning

TL;DR

This paper introduces a contact-aware motion planner for autonomous navigation of neuroendovascular tools, utilizing a kinematic model and pre-operative imaging to improve stroke treatment procedures.

Contribution

It presents a novel model-based planner that integrates anatomical geometry and tool mechanics for autonomous neuroendovascular navigation.

Findings

Achieved 100% success in navigating to the LCCA in over 50 trials.

Demonstrated robustness to aorta motion up to 10° and 10mm displacement.

Validated the approach with experimental navigation from the aorta to LCCA.

Abstract

In this paper, we propose a model-based contact-aware motion planner for autonomous navigation of neuroendovascular tools in acute ischemic stroke. The planner is designed to find the optimal control strategy for telescopic pre-bent catheterization tools such as guidewire and catheters, currently used for neuroendovascular procedures. A kinematic model for the telescoping tools and their interaction with the surrounding anatomy is derived to predict tools steering. By leveraging geometrical knowledge of the anatomy, obtained from pre-operative segmented 3D images, and the mechanics of the telescoping tools, the planner finds paths to the target enabled by interacting with the surroundings. We propose an actuation platform for insertion and rotation of the telescopic tools and present experimental results for the navigation from the base of the descending aorta to the LCCA. We demonstrate that, by leveraging the pre-operative plan, we can consistently navigate the LCCA with 100% success of over 50 independent trials. We also study the robustness of the planner towards motion of the aorta and errors in the initial positioning of the robotic tools. The proposed plan can successfully reach the LCCA for rotations of the aorta of up to 10{\deg}, and displacement of up to 10mm, on the coronal plane.